requestSort(column.id)}> {column.tooltip ? {column.label} : {column.label}} {getSortIndicator(column.id)}
{column.id === 'name' && isPricing && model.subRows && model.subRows.some(subRow => subRow.hasOwnProperty(!isMaxMode ? 'requests' : 'tokenInput')) ? {column.render ? column.render(model) : model[column.id]} {model.subRows.map(subRow => {subRow.name} {subRow.notes?.map((note, index) => )} )} : isPricing && model.subRows && model.subRows.some(subRow => subRow.hasOwnProperty(!isMaxMode ? 'requests' : 'tokenInput')) ? {column.render ? column.render(model) : model[column.id]} {model.subRows.map(subRow => {column.render ? column.render(subRow) : subRow[column.id]} )} : column.render ? column.render(model) : model[column.id]}

{error && <div className="absolute bottom-5 px-2 py-1 text-red-500 rounded-lg left-3 bg-red-100 text-sm animate-in fade-in-0 slide-in-from-bottom-1 duration-200">
            {error}
          </div>}
      </div>
      <div className="flex flex-col sm:flex-row sm:items-center sm:justify-between gap-2 py-2 px-4 pr-2 border border-neutral-200 dark:border-neutral-500 rounded-lg transition-all duration-200">
        <div className="flex items-center gap-2 font-medium capitalize transition-colors duration-200">
          {extractedServerName.length > 0 ? <span className="animate-in fade-in-0 slide-in-from-left-2 duration-300">
              {extractedServerName}
            </span> : <span className="text-neutral-300">No server detected</span>}
        </div>
        <div className="flex gap-2">
          <div className="relative">
            <button onClick={copyDeepLink} disabled={!extractedServerName} className="border border-neutral-300 dark:border-neutral-600 text-neutral-700 dark:text-neutral-300 py-1 px-2 rounded-lg hover:bg-neutral-100 dark:hover:bg-neutral-800 bg-white dark:bg-neutral-900 text-sm transition-all duration-200 disabled:opacity-50 disabled:cursor-not-allowed">
              Copy deeplink
            </button>
            {showOverlay === 'link' && <div className="absolute inset-0 border border-neutral-300 bg-neutral-100 dark:bg-neutral-800 rounded-lg flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200">
                <Icon icon="check" size={16} />
                <span className="text-sm">Copied</span>
              </div>}
          </div>
          <div className="relative">
            <button onClick={copyWebLink} disabled={!extractedServerName} className="border border-neutral-300 dark:border-neutral-600 text-neutral-700 dark:text-neutral-300 py-1 px-2 rounded-lg hover:bg-neutral-100 dark:hover:bg-neutral-800 bg-white dark:bg-neutral-900 text-sm transition-all duration-200 disabled:opacity-50 disabled:cursor-not-allowed">
              Copy web link
            </button>
            {showOverlay === 'weblink' && <div className="absolute inset-0 border border-neutral-300 bg-neutral-100 dark:bg-neutral-800 rounded-lg flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200">
                <Icon icon="check" size={16} />
                <span className="text-sm">Copied</span>
              </div>}
          </div>
        </div>
      </div>
      <div className="flex flex-col gap-2 p-1">
        <div className="flex flex-col gap-2">
          <Tabs>
            <Tab title="Markdown">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyMarkdownLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 w-full max-w-40 sm:max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark-md' && <div className="top-2 right-2 bottom-2 left-2 absolute inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyMarkdownLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 w-full max-w-40 sm:max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light-md' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in README</p>
            </Tab>
            <Tab title="HTML">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyHtmlLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyHtmlLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in README</p>
            </Tab>
            <Tab title="JSX">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyJsxLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark-jsx' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyJsxLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light-jsx' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in JSX components</p>
            </Tab>

</Tabs>
        </div>
      </div>
    </div>;
};

<div id="mcp-servers">
  # MCP-Server
</div>

MCP-Server kannst du per Cursor-Deeplinks installieren. Sie verwenden dasselbe Format wie [`mcp.json`](/de/context/model-context-protocol.mdx) mit einem Namen und einer Transportkonfiguration.

Installationslinks:

```
cursor://anysphere.cursor-deeplink/mcp/install?name=$NAME&config=$BASE64_ENCODED_CONFIG
```

<div id="generate-install-link">
  ## Installationslink erstellen
</div>

1. Namen und JSON-Konfiguration des Servers abrufen
2. Die Konfiguration mit `JSON.stringify` serialisieren und anschließend base64-codieren
3. `$NAME` und `$BASE64_ENCODED_CONFIG` durch den Namen und die codierte Konfiguration ersetzen

Helfer zum Generieren von Links:

<div id="example">
  ## Beispiel
</div>

Probier dieses JSON im MCP-Install-Link-Generator aus:

```json Konfiguration eines einzelnen MCP-Servers theme={null}
{
  "postgres": {
    "command": "npx",
    "args": [
      "-y",
      "@modelcontextprotocol/server-postgres",
      "postgresql://localhost/mydb"
    ]
  }
}
```

Ergebnis:

<div className="full-width-table mcp-install-examples">
  | Format       | Beispiel                                                                                                                                                                                                                                                                                                                                                              |
  | ------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
  | Text link    | [cursor://anysphere.curs...](cursor://anysphere.cursor-deeplink/mcp/install?name=postgres\&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=)                                                                                                                           |
  | Dark button  | <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=postgres&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=">  <img src="https://cursor.com/deeplink/mcp-install-dark.png" alt="Postgres-MCP-Server zu Cursor hinzufügen" style={{ maxHeight: 32 }} /></a>  |
  | Light button | <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=postgres&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=">  <img src="https://cursor.com/deeplink/mcp-install-light.png" alt="Postgres-MCP-Server zu Cursor hinzufügen" style={{ maxHeight: 32 }} /></a> |
</div>

<div id="install-server">
  ## Server installieren
</div>

1. Klicke auf den Link oder füge ihn in die Adressleiste deines Browsers ein
2. Cursor fordert dich auf, den Server zu installieren
3. Nutze den Server in Cursor

# MCP-Server
Source: https://docs.cursor.com/de/tools/mcp

Entdecke und installiere MCP-Server für Cursor

export const McpInstallButton = ({server, showIcon = true, prompt = null}) => {
  const [showModal, setShowModal] = useState(false);
  const InstallModal = ({isOpen, onClose, deepLink, server, children}) => {
    useEffect(() => {
      const handleKeyDown = event => {
        if (event.key === 'Escape') {
          onClose();
        }
      };
      if (isOpen) {
        document.addEventListener('keydown', handleKeyDown);
      }
      return () => {
        document.removeEventListener('keydown', handleKeyDown);
      };
    }, [isOpen, onClose]);
    if (!isOpen) return null;
    return <div className="fixed inset-0 bg-black bg-opacity-50 flex items-center justify-center z-50 transition-opacity duration-200" onClick={onClose}>
        <div className="bg-white dark:bg-neutral-900 rounded-lg p-6 max-w-md w-full mx-4 border border-neutral-200 dark:border-neutral-700 transition-all duration-200 transform" onClick={e => e.stopPropagation()}>
          <div className="mb-4">
            <h3 className="text-lg font-semibold text-black dark:text-white m-0 mb-2">
              Note
            </h3>
            <div className="text-neutral-600 dark:text-neutral-400">
              {children}
            </div>
          </div>
          <div className="flex gap-2 justify-end">
            <button onClick={onClose} className="px-3 py-1.5 text-xs font-medium rounded-md transition-colors duration-200 text-neutral-600 dark:text-neutral-400 hover:text-black dark:hover:text-white border border-neutral-200 dark:border-neutral-700 hover:bg-neutral-100 dark:hover:bg-neutral-800">
              Cancel
            </button>
            <a href={deepLink} onClick={onClose} target="_blank" className="px-3 py-1.5 text-xs font-medium rounded-md transition-colors duration-200 bg-black text-white hover:bg-neutral-800 dark:bg-neutral-800 dark:text-white dark:hover:bg-neutral-700 inline-flex items-center justify-center no-underline border-none !border-none">
              Continue
            </a>
          </div>
        </div>
      </div>;
  };
  const generateDeepLink = () => {
    if (!server || !server.name || !server.install) {
      return null;
    }
    try {
      if (typeof server.install === 'string') {
        return server.install;
      }
      if (server.install.url) {
        const config = {
          ...server.install
        };
        const jsonString = JSON.stringify(config);
        const utf8Bytes = new TextEncoder().encode(jsonString);
        const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
        const safeBase64Config = base64Config.replace(/\+/g, '%2B');
        return `cursor://anysphere.cursor-deeplink/mcp/install?name=${encodeURIComponent(server.name)}&config=${encodeURIComponent(safeBase64Config)}`;
      }
      if (server.install.command) {
        let config = {
          command: server.install.command,
          ...server.install.args && ({
            args: server.install.args
          }),
          ...server.install.env && ({
            env: server.install.env
          })
        };
        if (config.command && config.args) {
          const argsString = config.args.join(" ");
          config.command = `${config.command} ${argsString}`;
          delete config.args;
        }
        const jsonString = JSON.stringify(config);
        const utf8Bytes = new TextEncoder().encode(jsonString);
        const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
        const safeBase64Config = base64Config.replace(/\+/g, '%2B');
        return `cursor://anysphere.cursor-deeplink/mcp/install?name=${encodeURIComponent(server.name)}&config=${encodeURIComponent(safeBase64Config)}`;
      }
      return null;
    } catch (e) {
      console.error("Error generating deep link:", e);
      return null;
    }
  };
  const handleButtonClick = () => {
    setShowModal(true);
  };
  const handleClose = () => {
    setShowModal(false);
  };
  const deepLink = generateDeepLink();
  const isDocumentationOnly = typeof server?.install === 'string';
  const hasConfirmation = prompt || isDocumentationOnly;
  return <>
      {hasConfirmation ? <button onClick={handleButtonClick} className="inline-flex justify-center items-center gap-2 px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 not-prose text-black dark:text-white bg-white dark:bg-black border border-neutral-200 dark:border-neutral-700 hover:bg-neutral-100 dark:hover:bg-neutral-800">
          {showIcon && <Icon icon="plus" size={16} color="currentColor" />}
          Add to Cursor
        </button> : <a href={deepLink} className="inline-flex justify-center items-center gap-2 px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 not-prose text-black dark:text-white bg-white dark:bg-black border border-neutral-200 dark:border-neutral-700 hover:bg-neutral-100 dark:hover:bg-neutral-800">
          {showIcon && <Icon icon="plus" size={16} color="currentColor" />}
          Add to Cursor
        </a>}

{hasConfirmation && <InstallModal isOpen={showModal} onClose={handleClose} deepLink={deepLink} server={server}>
          {prompt}
        </InstallModal>}
    </>;
};

Erkunde und installiere MCP-Server für Cursor
Entwickler können [einen „Add to Cursor“-Button erstellen](/de/tools/developers#generate-install-link) und einen Link einbinden, den du in deiner Entwicklerdoku verwenden kannst, um die Installation in Cursor zu erleichtern

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Notion</h2>

<p className="m-0">All-in-one-Workspace für Notizen, Doks und Projektmanagement.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Figma</h2>

<p className="m-0">Design- und Collaboration-Plattform für Teams.</p>
    </div>

<McpInstallButton
      full
      server={{"name":"Figma","install":{"url":"http://127.0.0.1:3845/mcp"}}}
      prompt={<>
Lade die neueste Version der <a href="https://www.figma.com/downloads/" target="_blank">Figma-Desktop-App</a> herunter, um den Figma-MCP-Server zu hosten. Schau dir die offizielle <a href="https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Dev-Mode-MCP-Server" target="_blank">Figma-Setup-Anleitung</a> für Installationsschritte an
</>}
    />
  </Card>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Linear</h2>

<p className="m-0">Issue-Tracking und Projektmanagement für Developer-Teams.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">GitHub</h2>

<p className="m-0">Plattform für Versionskontrolle und kollaborative Entwicklung.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Playwright</h2>

<p className="m-0">End-to-End-Tests im Browser.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Sentry</h2>

<p className="m-0">Fehler-Tracking und Performance-Monitoring.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">DuckDB</h2>

<p className="m-0">In-Prozess-SQL-OLAP-Datenbank für lokale Analysen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Vercel</h2>

<p className="m-0">Projekte und Deployments auf Vercel managen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">GitLab</h2>

<p className="m-0">DevSecOps‑Plattform für Code, CI/CD und Sicherheit.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Atlassian</h2>

<p className="m-0">Tools für Projektmanagement und Zusammenarbeit, darunter Jira und Confluence.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PostHog</h2>

<p className="m-0">Analytics, Fehler-Tracking und Feature Flags.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Stripe</h2>

<p className="m-0">APIs für Zahlungsabwicklung.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PayPal</h2>

<p className="m-0">Payment-APIs.</p>
    </div>

<p className="m-0">dbt CLI, Semantic Layer und Discovery API.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Browserbase</h2>

<p className="m-0">Headless-Browser-Sessions für Agents.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Netlify</h2>

<p className="m-0">Webprojekte erstellen und deployen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Shopify</h2>

<p className="m-0">Entwicklungstools für Shopify-Apps.</p>
    </div>

<p className="m-0">Schwachstellenscan deiner Codebasis.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Zapier</h2>

<p className="m-0">Workflows zwischen Apps automatisieren.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Heroku</h2>

<p className="m-0">Verwalte Heroku-Apps und Ressourcen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Hugging Face</h2>

<p className="m-0">Greif auf den Hugging Face Hub und die Gradio MCP-Server zu.</p>
    </div>

<p className="m-0">Wix-Websites erstellen und verwalten.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Semgrep</h2>

<p className="m-0">Code auf Sicherheitslücken prüfen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Plaid</h2>

<p className="m-0">Greif auf Finanzkontodaten zu.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Mercado Pago</h2>

<p className="m-0">Greif auf die Mercado-Pago-Dokumentation zu.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">LaunchDarkly</h2>

<p className="m-0">Feature Flags als Service.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Context7</h2>

<p className="m-0">Aktualisierte Code-Dokumentation.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">AWS-Dokumentation</h2>

<p className="m-0">
        Greif auf die AWS-Dokumentation zu, such und hol dir Empfehlungen.
      </p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Endor Labs</h2>

<p className="m-0">Security-Risiko-Insights für Code.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">InstantDB</h2>

<p className="m-0">InstantDB abfragen und verwalten.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Tinybird</h2>

<p className="m-0">APIs für Echtzeit-Analytics.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">MongoDB</h2>

<p className="m-0">MongoDB-Daten und Deployments verwalten.</p>
    </div>

<p className="m-0">Verwalte Neon Postgres.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">SonarQube</h2>

<p className="m-0">Code mit SonarQube analysieren.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Socket</h2>

<p className="m-0">Abhängigkeiten analysieren und absichern.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Select Star</h2>

<p className="m-0">Datenkatalog, Datenherkunft und Kontext.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Pipedream</h2>

<p className="m-0">Mit APIs und Workflows verbinden.</p>
    </div>

<p className="m-0">Auth0-Ressourcen verwalten.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">MS Learn Docs</h2>

<p className="m-0">Durchsuche die Microsoft-Dokumentation.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Mercado Libre</h2>

<p className="m-0">Greif auf die Mercado‑Libre-Dokumentation zu.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Honeycomb</h2>

<p className="m-0">Observability-Daten und SLOs abfragen</p>
    </div>

<Card>
    <div>
      <div className="size-7 mb-3 text-neutral-900 dark:text-white aspect-square server-icon">
        <svg xmlns="http://www.w3.org/2000/svg" xmlnsXlink="http://www.w3.org/1999/xlink" width="auto" height="100%" viewBox="0 0 1080 1080" xmlSpace="preserve">
          <desc>
            {"Erstellt mit Fabric.js 5.2.4"}
          </desc>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PagerDuty</h2>

<p className="m-0">Incidents und Alerts verwalten.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Graphite</h2>

<p className="m-0">Erstelle und verwalte gestapelte PRs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Railway</h2>

<p className="m-0">Apps, Datenbanken und Services deployen.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Postman</h2>

<p className="m-0">API-Zusammenarbeit und Tests.</p>
    </div>

<div className="text-left mb-16 mt-4 max-w-[52rem] mx-auto flex flex-col gap-3 px-4 items-start text-sm">
  <p className="text-neutral-600 dark:text-neutral-400">
    Das ist eine kuratierte Auswahl von MCP‑Tools für Entwickler von offiziellen Anbietern.
    Wenn deine Firma oder dein Tool einen offiziellen MCP‑Server bereitstellt und du ihn
    aufgenommen sehen willst, dann{" "}

<a target="_blank" href="https://github.com/cursor/mcp-servers" className="text-black dark:text-white inline underline decoration-neutral-200 underline-offset-4 font-medium hover:decoration-neutral-400 transition-colors">
      erstelle auf GitHub einen PR oder ein Issue
    </a>

{" "}

und wir prüfen die Aufnahme.
  </p>
</div>

# Häufige Probleme
Source: https://docs.cursor.com/de/troubleshooting/common-issues

Lösungen für häufige Probleme und FAQs

Unten findest du häufige Probleme und ihre Lösungen.

<div id="networking-issues">
  ### Netzwerkprobleme
</div>

Check zuerst deine Netzwerkverbindung. Geh zu `Cursor Settings` > `Network` und klick auf `Run Diagnostics`. Das testet deine Verbindung zu den Cursor‑Servern und hilft, netzwerkbedingte Probleme zu erkennen, die AI‑Features, Updates oder andere Online‑Funktionen beeinträchtigen könnten.

Cursor setzt für AI‑Features auf HTTP/2, weil damit gestreamte Antworten effizient übertragen werden. Wenn dein Netzwerk kein HTTP/2 unterstützt, kann es zu Indexierungsfehlern und Problemen mit AI‑Features kommen.

Das passiert oft in Unternehmensnetzwerken, mit VPNs oder bei Proxys wie Zscaler.

Um das zu beheben, aktiviere das HTTP/1.1‑Fallback in den App‑Einstellungen (nicht in den Cursor‑Einstellungen): Drück `CMD/CTRL + ,`, such nach `HTTP/2` und aktiviere `Disable HTTP/2`. Dadurch wird HTTP/1.1 erzwungen und das Problem behoben.

Wir planen, automatische Erkennung und automatisches Fallback hinzuzufügen.

<div id="resource-issues-cpu-ram-etc">
  ### Ressourcenprobleme (CPU, RAM, etc.)
</div>

Hohe CPU‑ oder RAM‑Auslastung kann deinen Rechner verlangsamen oder Ressourcenwarnungen auslösen.

Große Codebasen benötigen zwar mehr Ressourcen, aber hohe Auslastung kommt meist von Erweiterungen oder fehlerhaften Einstellungen.

<Note>
  Wenn du auf **macOS** eine Warnung für wenig RAM siehst, beachte bitte, dass es bei einigen Nutzer:innen einen Bug gibt, der völlig falsche Werte anzeigen kann. Öffne in diesem Fall den Aktivitätsmonitor und schau im Tab „Memory“ nach, um die korrekte Speicherauslastung zu sehen.
</Note>

Wenn du hohe CPU‑ oder RAM‑Auslastung feststellst, probier Folgendes:

<AccordionGroup>
  <Accordion title="Check deine Erweiterungen">
    Erweiterungen können die Performance beeinflussen.

Der Extension Monitor zeigt den Ressourcenverbrauch aller installierten und integrierten Erweiterungen.

Aktiviere den Extension Monitor unter `Settings` > `Application` > `Experimental` und schalte `Extension Monitor: Enabled` ein. Danach wirst du aufgefordert, Cursor neu zu starten.

Öffnen: `Cmd/Ctrl + Shift + P` → `Developer: Open Extension Monitor`.

Cursor führt deine Erweiterungen in einem oder mehreren **Extension Hosts** aus. Meist laufen die meisten deiner Erweiterungen im selben Extension Host – eine Erweiterung mit hohem CPU‑Verbrauch kann dadurch ihre Nachbarerweiterungen ausbremsen!

Der Extension Monitor zeigt:

* Jeden langlebigen Prozess, der von einer Erweiterung gestartet wurde (nur macOS und Linux).
    * **% Ext Host**: Den Prozentsatz der gesamten Extension‑Host‑Zeit, den diese Erweiterung verbraucht. Hilft zu erkennen, welche Erweiterungen relativ am meisten Zeit nutzen.
    * **Max Blocking**: Den längsten ununterbrochenen Ausführungsblock einer Erweiterung pro Monitoring‑Intervall.
    * **% CPU**:
      * Für Erweiterungen: Den Prozentsatz der gesamten CPU‑Nutzung, der dem Code der Erweiterung zugerechnet wird.
      * Für Prozesse: Den Prozentsatz der gesamten CPU‑Nutzung, der dem gestarteten Prozess zugerechnet wird (nur macOS und Linux).
    * **Memory**:
      * Für Erweiterungen: Die Menge an JS‑Heap‑Speicher, die vom Code der Erweiterung genutzt wird (externe Allokationen nicht enthalten).
      * Für Prozesse: Die Menge an Systemspeicher, die vom gestarteten Prozess genutzt wird (nur macOS und Linux).

Du kannst auch testen, indem du `cursor --disable-extensions` in der Kommandozeile ausführst. Wenn sich die Performance verbessert, aktiviere Erweiterungen nacheinander wieder, um die problematischen zu finden.

Probier Extension Bisect, um problematische Erweiterungen zu identifizieren. Mehr dazu [hier](https://code.visualstudio.com/blogs/2021/02/16/extension-bisect#_welcome-extension-bisect). Hinweis: Das funktioniert am besten bei sofort auftretenden Problemen, nicht bei schleichender Performance‑Verschlechterung.
  </Accordion>

<Accordion title="Verwende den Process Explorer">
    Der Process Explorer zeigt, welche Prozesse Ressourcen verbrauchen.

Öffnen: Command Palette (`Cmd/Ctrl + Shift + P`) → `Developer: Open Process Explorer`.

Check die Prozesse unter:

* **`extensionHost`**: Erweiterungsbezogene Probleme
    * **`ptyHost`**: Ressourcenverbrauch des Terminals

Der Process Explorer zeigt jedes Terminal und seine laufenden Befehle zur Diagnose an.

Für andere Prozesse mit hoher Auslastung melde dich im [Forum](https://forum.cursor.com/).
  </Accordion>

<Accordion title="Systemressourcen monitoren">
    Nutze die Monitoring‑Tools deines Betriebssystems, um festzustellen, ob das Problem Cursor‑spezifisch oder systemweit ist.
  </Accordion>

<Accordion title="Minimale Installation testen">
    Wenn Probleme weiterhin bestehen, teste eine minimale Cursor‑Installation.
  </Accordion>
</AccordionGroup>

<div id="general-faqs">
  ## Allgemeine FAQs
</div>

<AccordionGroup>
  <Accordion title="Ich sehe ein Update im Changelog, aber Cursor aktualisiert sich nicht">
    Neue Updates werden in Wellen ausgerollt – zufällig ausgewählte Nutzer bekommen sie zuerst. Dein Update sollte in ein paar Tagen kommen.
  </Accordion>

<Accordion title="Ich habe Probleme mit meinem GitHub-Login in Cursor / Wie logge ich mich in Cursor aus GitHub aus?">
    Nutze `Sign Out of GitHub` in der Befehlspalette `Ctrl/⌘ + Shift + P`.
  </Accordion>

<Accordion title="Ich kann GitHub Codespaces nicht verwenden">
    GitHub Codespaces wird noch nicht unterstützt.
  </Accordion>

<Accordion title="Ich habe Fehler bei der Verbindung zu Remote SSH">
    SSH zu Mac- oder Windows-Maschinen wird nicht unterstützt. Für andere Probleme poste mit Logs im [Forum](https://forum.cursor.com/).
  </Accordion>

<Accordion title="SSH-Verbindungsprobleme unter Windows">
    Wenn du „SSH is only supported in Microsoft versions of VS Code“ siehst:

1. Remote-SSH deinstallieren:
       * Erweiterungen-Ansicht öffnen (`Ctrl + Shift + X`)
       * Nach „Remote-SSH“ suchen
       * Zahnrad-Symbol klicken → „Uninstall“

2. Anysphere Remote SSH installieren:
       * Cursor-Marktplatz öffnen
       * Nach „SSH“ suchen
       * Die Anysphere Remote SSH-Erweiterung installieren

3. Nach der Installation:
       * Alle VS Code-Instanzen mit aktiven SSH-Verbindungen schließen
       * Cursor neu starten
       * Erneut per SSH verbinden

Prüfe, ob deine SSH-Konfiguration und -Keys korrekt eingerichtet sind.
  </Accordion>

<Accordion title="Cursor Tab und Inline Edit funktionieren hinter meinem Firmen-Proxy nicht">
    Cursor Tab und Inline Edit verwenden HTTP/2 für geringere Latenz und Ressourcennutzung. Einige Unternehmens-Proxys (z. B. Zscaler) blockieren HTTP/2. Beheb das, indem du `"cursor.general.disableHttp2": true` in den Einstellungen setzt (`Cmd/Ctrl + ,`, nach `http2` suchen).
  </Accordion>

<Accordion title="Ich habe gerade Pro abonniert, aber in der App bin ich immer noch im Free-Plan">
    Melde dich in den Cursor-Einstellungen ab und wieder an.
  </Accordion>

<Accordion title="Wann wird meine Nutzung wieder zurückgesetzt?">
    Pro-Abonnenten: Klicke im [Dashboard](https://cursor.com/dashboard) auf `Manage Subscription`, um dein Verlängerungsdatum zu sehen.

Free-User: Check das Datum deiner ersten Cursor-E-Mail. Die Nutzung wird monatlich ab diesem Datum zurückgesetzt.
  </Accordion>

<Accordion title="Meine Chat-/Composer-Historie ist nach einem Update verschwunden">
    Wenig Speicherplatz kann dazu führen, dass Cursor während Updates historische Daten löscht. Um das zu verhindern:

1. Vor dem Update genug freien Speicherplatz sicherstellen
    2. Regelmäßig unnötige Systemdateien bereinigen
    3. Wichtige Unterhaltungen vor dem Update sichern
  </Accordion>

<Accordion title="Wie deinstalliere ich Cursor?">
    Folge [dieser Anleitung](https://code.visualstudio.com/docs/setup/uninstall). Ersetze „VS Code“ oder „Code“ durch „Cursor“ und „.vscode“ durch „.cursor“.
  </Accordion>

<Accordion title="Wie lösche ich mein Konto?">
    Klicke im [Dashboard](https://cursor.com/dashboard) auf `Delete Account`. Dadurch werden dein Konto und alle zugehörigen Daten dauerhaft gelöscht.
  </Accordion>

<Accordion title="Wie öffne ich Cursor über die Kommandozeile?">
    Führe `cursor` in deinem Terminal aus. Falls der Befehl fehlt:

1. Befehlspalette öffnen `⌘⇧P`
    2. `install command` eingeben
    3. `Install 'cursor' command` auswählen (optional den `code`-Befehl installieren, um den von VS Code zu überschreiben)
  </Accordion>

<Accordion title="Anmeldung bei Cursor nicht möglich">
    Wenn ein Klick auf Sign In zu cursor.com weiterleitet, dich aber nicht anmeldet, deaktivier deine Firewall oder Antiviren-Software – die könnte den Anmeldeprozess blockieren.
  </Accordion>

<Accordion title="Meldung über verdächtige Aktivität">
    Aufgrund der jüngsten Zunahme von Missbrauch könnte deine Anfrage aus Sicherheitsgründen blockiert worden sein. So löst du das:

Check zuerst dein VPN. Wenn du eins nutzt, probier es auszuschalten, da VPNs manchmal unsere Sicherheitssysteme triggern.

Wenn das nicht hilft, kannst du Folgendes probieren:

* Einen neuen Chat erstellen
    * Kurz warten und später erneut versuchen
    * Ein neues Konto mit Google- oder GitHub-Authentifizierung erstellen
    * Auf Cursor Pro upgraden
  </Accordion>
</AccordionGroup>

# Eine Request-ID abrufen
Source: https://docs.cursor.com/de/troubleshooting/request-reporting

Request-IDs für den technischen Support finden

Wenn das Cursor-Team ein technisches Problem untersucht, bitten wir dich möglicherweise um eine „Request-ID“.

<div id="what-is-a-request-id">
  ## Was ist eine Request-ID?
</div>

Eine Request-ID dient zur eindeutigen Identifizierung jeder Anfrage an Cursor in unseren internen Systemen.

Formatbeispiel: `8f2a5b91-4d3e-47c6-9f12-5e8d94ca7d23`

<div id="how-do-i-find-a-request-id">
  ## Wie finde ich eine Request-ID?
</div>

<Warning>
  Request-IDs sind eingeschränkt, wenn der Privacy Mode aktiviert ist. Deaktiviere den Privacy Mode, wenn du Probleme meldest.

Hinweis: Bei Nutzer:innen des Business-Plans ist der Privacy Mode standardmäßig vom Admin der Organisation aktiviert.
</Warning>

<div id="getting-your-current-request-id">
  ### Deine aktuelle Request-ID abrufen
</div>

So meldest du ein Problem in deinem aktuellen oder jüngsten Chat:

Öffne die Unterhaltung in der Chat-Sidebar, öffne das Kontextmenü (oben rechts) und wähle `Copy Request ID`.

Schick uns die kopierte Request-ID wie angefordert über das Forum oder per E-Mail.

<div id="getting-a-request-id-from-a-previous-action">
  ### Eine Request-ID aus einer früheren Aktion abrufen
</div>

Ruf frühere Request-IDs mit dem Befehl `Report AI Action` ab:

1. Öffne die Command Palette `⌘⇧P`
2. Tippe `Report AI Action`
3. Wähle die Option `Report AI Action` aus

Dadurch werden deine letzten AI-Aktionen in Chat, CMD+K und Apply angezeigt.

Wähle die Aktion anhand von Zeitpunkt und Feature aus. Kopiere die Request-ID und schick sie uns.

# Leitfaden zur Fehlerbehebung
Source: https://docs.cursor.com/de/troubleshooting/troubleshooting-guide

Schritte zur Behebung von Problemen und zum Melden von Bugs

Probleme mit Cursor können durch Extensions, App-Daten oder Systemprobleme verursacht werden. Probier diese Schritte zur Fehlerbehebung aus.

<CardGroup cols={1}>
  <Card horizontal title="Ein Problem melden" icon="bug" href="#reporting-an-issue">
    Schritte, um ein Problem an das Cursor-Team zu melden
  </Card>
</CardGroup>

<div id="troubleshooting">
  ## Fehlerbehebung
</div>

<Steps>
  <Step title="Netzwerkverbindung prüfen">
    Check zuerst, ob Cursor seine Dienste erreichen kann.

**Netzwerkdiagnose ausführen:** Geh zu `Cursor Settings` > `Network` und klick auf `Run Diagnostics`. Das testet deine Verbindung zu den Cursor-Servern und erkennt Netzwerkprobleme, die KI-Features, Updates oder andere Online-Funktionen beeinträchtigen.

Wenn die Diagnose Verbindungsprobleme anzeigt, check Firewall-Einstellungen, Proxy-Konfiguration oder Netzwerkeinschränkungen, die den Zugriff von Cursor blockieren.
  </Step>

<Step title="Erweiterungsdaten löschen">
    Bei Problemen mit Erweiterungen:

**Alle Erweiterungen temporär deaktivieren:** Führ `cursor --disable-extensions` in der Kommandozeile aus. Wenn die Probleme weg sind, aktiviere die Erweiterungen nacheinander wieder, um die fehlerhafte zu finden.

**Erweiterungsdaten zurücksetzen:** Deinstallier und installier problematische Erweiterungen neu, um ihre gespeicherten Daten zurückzusetzen. Check die Einstellungen auf Erweiterungskonfigurationen, die nach der Neuinstallation bestehen bleiben.
  </Step>

<Step title="App-Daten löschen">
    <Warning>
      Dadurch werden deine App-Daten gelöscht, einschließlich Erweiterungen, Themes, Snippets und installationsbezogener Daten. Exportier zuerst dein Profil, um diese Daten zu sichern.
    </Warning>

Cursor speichert App-Daten außerhalb der App, damit sie über Updates und Neuinstallationen hinweg wiederhergestellt werden können.

So löschst du App-Daten:

**Windows:** Führ diese Befehle in der Eingabeaufforderung aus:

```txt  theme={null}
    rd /s /q "%USERPROFILE%\AppData\Local\Programs\Cursor"
    rd /s /q "%USERPROFILE%\AppData\Local\Cursor"
    rd /s /q "%USERPROFILE%\AppData\Roaming\Cursor"
    del /f /q "%USERPROFILE%\.cursor*"
    rd /s /q "%USERPROFILE%\.cursor"
    ```

**macOS:** Führ `sudo rm -rf ~/Library/Application\ Support/Cursor` und `rm -f ~/.cursor.json` im Terminal aus.

**Linux:** Führ `rm -rf ~/.cursor ~/.config/Cursor/` im Terminal aus.
  </Step>

<Step title="Cursor deinstallieren">
    So deinstallierst du Cursor:

<CardGroup cols={1}>
      <Card horizontal title="Windows" icon="windows">
        Such im Startmenü nach „Add or Remove Programs“, finde „Cursor“, klick auf „Uninstall“.
      </Card>

<Card horizontal title="macOS" icon="apple">
        Öffne den Programme-Ordner, klick mit Rechts auf „Cursor“, wähl „Move to Trash“.
      </Card>

<Card horizontal title="Linux" icon="linux">
        **Für .deb-Pakete:** `sudo apt remove cursor`

**Für .rpm-Pakete:** `sudo dnf remove cursor` oder `sudo yum remove cursor`

**Für AppImage:** Lösch die Datei Cursor.appimage an ihrem Speicherort.
      </Card>
    </CardGroup>
  </Step>

<Step title="Cursor neu installieren">
    Installier neu über die [Download-Seite](https://www.cursor.com/downloads). Wenn du die App-Daten nicht gelöscht hast, stellt Cursor den vorherigen Zustand wieder her. Andernfalls bekommst du eine frische Installation.
  </Step>
</Steps>

<div id="reporting-an-issue">
  ## Ein Problem melden
</div>

Wenn diese Schritte nicht helfen, melde dich im [Forum](https://forum.cursor.com/).

<Card horizontal title="Cursor-Forum" icon="message" href="https://forum.cursor.com/">
  Melde einen Bug oder ein Problem im Cursor-Forum
</Card>

Für eine schnelle Lösung gib Folgendes an:

<CardGroup cols={2}>
  <Card title="Screenshot des Problems" icon="image">
    Mach einen Screenshot und schwärze sensible Informationen.
  </Card>

<Card title="Schritte zur Reproduktion" icon="list-check">
    Dokumentiere die genauen Schritte, um das Problem zu reproduzieren.
  </Card>

<Card title="Systeminformationen" icon="computer">
    Hol dir die Systeminfos von:

<br />

`Cursor` > `Help` > `About`
  </Card>

<Card title="Request-IDs" icon="shield-halved" href="/de/troubleshooting/request-reporting">
    Klicke, um unseren Guide zum Sammeln von Request-IDs zu öffnen
  </Card>

<Card title="Konsolenfehler" icon="bug">
    Öffne die Entwicklerkonsole: <br />
    `Developer: Toggle Developer Tools`
  </Card>

<Card title="Logs" icon="file-lines">
    Öffne die Logs: <br />
    `Developer: Open Logs Folder`
  </Card>
</CardGroup>

# Willkommen
Source: https://docs.cursor.com/de/welcome

Erfahre mehr über Cursor und wie du loslegst

Cursor ist ein KI-gestützter Code-Editor, der deine Codebase versteht und dir mit natürlicher Sprache hilft, schneller zu coden. Beschreib einfach, was du bauen oder ändern willst, und Cursor generiert den Code für dich.

<CardGroup cols={3}>
  <Card title="Loslegen" icon="rocket" href="/de/get-started/installation">
    <div>
      Lade Cursor herunter, installier es und leg in wenigen Minuten los
    </div>
  </Card>

<Card title="Changelog" icon="sparkles" href="https://www.cursor.com/changelog">
    <div>
      Bleib über die neuesten Features und Verbesserungen auf dem Laufenden
    </div>
  </Card>

<Card title="CLI" icon="terminal" href="/de/cli/overview">
    <div>
      Nutze Cursor in deinem Terminal
    </div>
  </Card>

<Card title="Konzepte" icon="lightbulb" href="/de/get-started/concepts">
    <div>
      Versteh die Kernkonzepte und Features, die Cursor antreiben
    </div>
  </Card>

<Card title="Modelle" icon="brain" href="/de/models">
    <div>
      Entdecke verfügbare KI-Modelle und wie du das richtige auswählst
    </div>
  </Card>

<Card title="Guides" icon="book" href="/de/guides/working-with-context">
    <div>
      Lern Best Practices und Workflows für verschiedene Use Cases
    </div>
  </Card>

<Card title="Downloads" icon="download" href="https://cursor.com/downloads" arrow>
    <div>
      Hol dir Cursor für deinen Computer
    </div>
  </Card>

<Card title="Forum" icon="message" href="https://forum.cursor.com">
    <div>
      Für technische Fragen und zum Erfahrungsaustausch, besuch unser Forum
    </div>
  </Card>

<Card title="Support" icon="headset" href="mailto:hi@cursor.com">
    <div>
      Bei Konto- und Abrechnungsfragen schreib unserem Support-Team eine E‑Mail
    </div>
  </Card>
</CardGroup>

# Agent Security
Source: https://docs.cursor.com/en/account/agent-security

Security considerations for using Cursor Agent

Prompt injection, AI hallucinations, and other issues may cause AI to behave in unexpected and potentially malicious ways. While we continue to work on solving prompt injection at a more foundational level, our primary protection in Cursor products are guardrails around what an agent can do, including requiring manual approval for sensitive actions by default. The goal of this document is to explain our guardrails and what users can expect from them.

All of the controls and behavior below are our default and recommended settings.

## First-party tool calls

Cursor comes bundled with tools that enable the agent to help our users write code. These include file read, edits, calling terminal commands, searching the web for documentation, and others.

Read tools do not require approval (ie reading files, searching across code). Users may use [.cursorignore](/en/context/ignore-files) to block the agent from accessing specific files at all, but otherwise reads are generally permitted without approval. For actions that carry risk of exfiltrating sensitive data, we require explicit approval.

Modifying files within the current workspace does not require explicit approval with some exceptions. When an agent makes changes to files they are immediately saved to disk. We recommend that Cursor be run in version controlled workspaces, so that the contents of the files can be reverted at any time. We require explicit approval before changing files that modify the configuration of our IDE/CLI, such as the workspace settings file of the editor. However, users automatically reloading on file change should be aware that agent changes to files may trigger automatic execution before they have had a chance to review the changes.

Any terminal command suggested by Agents requires approval by default. We recommend that users review every command before the agent executes it. Users who accept the risk may choose to enable the agent to run all commands without approval. We include an [allowlist](/en/agent/tools) feature in Cursor, but do not consider it to be a security control. Some users choose to allow specific commands, but this is a best effort system and bypasses may be possible. We do not recommend "Run Everything", which bypasses any configured allowlists.

## Third-party tool calls

Cursor enables connecting external tools via [MCP](/en/context/mcp). All third party MCP connections must be explicitly approved by the user. Once a user approves an MCP, by default each tool call suggested in Agent Mode for every external MCP integration must be explicitly approved prior to execution.

## Network requests

Network requests may be used by an attacker to exfiltrate data. We presently do not support any first-party tools making network requests outside of a very select set of hosts (ie Github), explicit link retrieval, and to support web search with a select set of providers. Arbitrary agent network requests are prevented with default settings.

## Workspace trust

The Cursor IDE supports the standard [workspace trust](https://code.visualstudio.com/docs/editing/workspaces/workspace-trust) feature that is *disabled* by default. Workspace trust presents users with a prompt when they open a new workspace to choose normal or restricted mode. Restricted mode breaks AI and other features our users typically use Cursor for. We recommend other tools, such as a basic text editor for working with repos that you do not trust.

Workspace trust may be enabled in a users' settings by performing these steps:

1. Open your user settings.json file
2. Add the following configuration:
   ```json  theme={null}
   "security.workspace.trust.enabled": true
   ```

This setting can also be enforced organization-wide through Mobile Device Management (MDM) solutions.

## Responsible disclosure

If you believe you have found a vulnerability in Cursor, please follow the guide on our GitHub Security page and submit the report there. If you're unable to use GitHub, you may also reach us at [security@cursor.com](mailto:security@cursor.com).

We commit to acknowledging vulnerability reports within 5 business days, and addressing them as soon as we are able to. We will publish the results in the form of security advisories on our GitHub security page. Critical incidents will be communicated both on the GitHub security page and via email to all users.

# Billing
Source: https://docs.cursor.com/en/account/billing

Managing Cursor subscriptions, refunds, and invoices

### How do I access billing settings?

Access the billing portal through the [Dashboard](https://cursor.com/dashboard) by clicking "Billing" in your Dashboard. This opens a secure portal for all billing tasks.

### What are Cursor's billing cycles?

Billing cycles run monthly or annually, starting on your subscription date. Teams accounts are charged per seat with pro-rated billing for new members.

### How do seats work for Teams accounts?

Teams accounts charge per seat (one per Team member). When adding members mid-cycle, you're only charged for their remaining time. If a member has used any credits and is removed, their seat remains occupied until the end of the billing cycle - no pro-rated refunds are given. Team admins can manage seats through the dashboard.

### Can I switch between monthly and annual billing?

Yes! Here's how:

**Pro plan**

1. Go to the Cursor [dashboard](https://cursor.com/dashboard)
2. Click "Billing and Invoices" from the left sidebar to reach the billing page
3. Click "Manage subscription"
4. Click "Update subscription"
5. Select "Yearly" or "Monthly", then click "Continue"

**Teams plan**

1. Go to the Cursor [dashboard](https://cursor.com/dashboard)
2. Click "Billing and Invoices" from the left sidebar to reach the billing page
3. Click the "Upgrade Now" button to switch to yearly billing

<Note>
  You can only switch from monthly to yearly billing self-serve. To switch from yearly to monthly, contact us at
  [hi@cursor.com](mailto:hi@cursor.com).
</Note>

### Where can I find my invoices?

Find all billing history in the billing portal. View and download current and past invoices.

### Can I get invoices automatically emailed to me?

Invoices must be downloaded manually from the billing portal. We're developing automatic invoice emails. You'll be able to opt-in once available.

### How do I update my billing information?

Update payment method, company name, address, and tax information through the billing portal. We use Stripe for secure transactions. Changes only affect future invoices, we cannot modify historical invoices.

### How do I cancel my subscription?

Cancel your subscription through the Billing and Invoices page by clicking "Manage Subscription" the "Cancel subscription" button. Access continues until the end of your current billing period.

### I'm having other billing issues. How can I get help?

For billing questions not covered here, email [hi@cursor.com](mailto:hi@cursor.com) from the email linked to your account. Please include your account details and concerns.

# Pricing
Source: https://docs.cursor.com/en/account/pricing

Cursor's plans and their pricing

You can try Cursor for free or purchase an individual or team plan.

## Individual

All individual plans include:

* Unlimited tab completions
* Extended agent usage limits on all models
* Access to Bugbot
* Access to Background Agents

Each plan includes usage charged at model inference [API prices](/en/models#model-pricing):

* Pro includes \$20 of API agent usage + additional bonus usage
* Pro Plus includes \$70 of API agent usage + additional bonus usage
* Ultra includes \$400 of API agent usage + additional bonus usage

We work hard to grant additional bonus capacity beyond the guaranteed included usage. Since different models have different API costs, your model selection affects token output and how quickly your included usage is consumed. You can view usage and token breakdowns on [your dashboard](https://cursor.com/dashboard?tab=usage). Limit notifications are routinely shown in the editor.

### How much usage do I need?

Based on our usage data, you can expect the following usage levels:

* **Daily Tab users**: Always stay within \$20
* **Limited Agent users**: Often stay within the included \$20
* **Daily Agent users**: Typically \$60–\$100/mo total usage
* **Power users (multiple agents/automation)**: Often \$200+/mo total usage

Based on our usage data, limits are roughly equivalent to the following for a *median user*:

* Pro: \~225 Sonnet 4 requests, \~550 Gemini requests, or \~500 GPT 5 requests
* Pro+: \~675 Sonnet 4 requests, \~1,650 Gemini requests, or \~1,500 GPT 5 requests
* Ultra: \~4,500 Sonnet 4 requests, \~11,000 Gemini requests, or \~10,000 GPT 5 requests

### What happens when I reach my limit?

When you exceed your included monthly usage, you'll be notified in the editor and can choose to:

* **Add on-demand usage**: Continue using Cursor at the same API rates with pay-as-you-go billing
* **Upgrade your plan**: Move to a higher tier for more included usage

On-demand usage is billed monthly at the same rates as your included usage. Requests are never downgraded in quality or speed.

## Teams

There are two teams plans: Teams (\$40/user/mo) and Enterprise (Custom).

Team plans provide additional features like:

* Privacy Mode enforcement
* Admin Dashboard with usage stats
* Centralized team billing
* SAML/OIDC SSO

We recommend Teams for any customer that is happy self-serving. We recommend [Enterprise](/en/contact-sales) for customers that need priority support, pooled usage, invoicing, SCIM, or advanced security controls.

Learn more about [Teams pricing](/en/account/teams/pricing).

## Auto

Enabling Auto allows Cursor to select the premium model best fit for the immediate task and with the highest reliability based on current demand. This feature can detect degraded output performance and automatically switch models to resolve it.

<Note>We've heavily invested in the quality and overall performance of Auto. Starting at your next billing renewal after September 15, Auto consumes usage at the following API rates.</Note>

* **Input + Cache Write**: \$1.25 per 1M tokens
* **Output**: \$6.00 per 1M tokens
* **Cache Read**: \$0.25 per 1M tokens

Both the editor and dashboard will show your usage, which includes Auto. If you prefer to select a model directly, usage is incurred at that model's list API price.

## Max Mode

Certain models have the ability to use [Max Mode](/en/models#max-mode), which allows for longer reasoning and larger context windows up to 1M tokens. While the majority of coding tasks do not need to use Max Mode, it can be helpful for more complex queries, especially with large files or codebases. Using Max Mode will consume more usage. You can view all requests and token breakdowns on [your dashboard](https://cursor.com/dashboard?tab=usage).

## Bugbot

Bugbot is a separate product from Cursor subscriptions and has its own pricing plan.

* **Pro** (\$40/mo): Unlimited reviews on up to 200 PRs/month, unlimited access to Cursor Ask, integration with Cursor to fix bugs, and access to Bugbot Rules
* **Teams** (\$40/user/mo): Unlimited code reviews across all PRs, unlimited access to Cursor Ask, pooled usage across your team, and advanced rules and settings
* **Enterprise** (Custom): Everything in Teams plus advanced analytics and reporting, priority support, and account management

Learn more about [Bugbot pricing](https://cursor.com/bugbot#pricing).

## Background Agent

Background Agents are charged at API pricing for the selected [model](/en/models). You'll be asked to set a spend limit for Background Agents when you first start using them.

<Info>
  Virtual Machine (VM) compute for background agents will be priced in the future.
</Info>

# Admin API
Source: https://docs.cursor.com/en/account/teams/admin-api

Access team metrics, usage data, and spending information via API

The Admin API lets you programmatically access your team's data, including member information, usage metrics, and spending details. Build custom dashboards, monitoring tools, or integrate with existing workflows.

<Note>
  The API is in its first release. We're expanding capabilities based on feedback - let us know what endpoints you need!
</Note>

## Authentication

All API requests require authentication using an API key. Only team administrators can create and manage API keys.

API keys are tied to the organization, viewable by all admins, and are unaffected by the original creator's account status.

### Creating an API Key

1. Navigate to **cursor.com/dashboard** → **Settings** tab → **Cursor Admin API Keys**
2. Click **Create New API Key**
3. Give your key a descriptive name (e.g., "Usage Dashboard Integration")
4. Copy the generated key immediately - you won't see it again

Format: `key_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx`

### Using Your API Key

Use your API key as the username in basic authentication:

**Using curl with basic auth:**

```bash  theme={null}
curl https://api.cursor.com/{route} -u API_KEY:
```

**Or set the Authorization header directly:**

```bash  theme={null}
Authorization: Basic {base64_encode('API_KEY:')}
```

## Base URL

All API endpoints use:

```
https://api.cursor.com
```

## Endpoints

### Get Team Members

Retrieve all team members and their details.

```
GET /teams/members
```

#### Response

Returns an array of team member objects:

```typescript  theme={null}
{
  teamMembers: {
    name: string;
    email: string;
    role: 'owner' | 'member' | 'free-owner';
  }[];
}
```

#### Example Response

```json  theme={null}
{
  "teamMembers": [
    {
      "name": "Alex",
      "email": "developer@company.com",
      "role": "member"
    },
    {
      "name": "Sam",
      "email": "admin@company.com",
      "role": "owner"
    }
  ]
}

```

#### Example Request

```bash  theme={null}
curl -X GET https://api.cursor.com/teams/members \
  -u YOUR_API_KEY:
```

### Get Daily Usage Data

Retrieve detailed daily usage metrics for your team within a date range. Provides insights into code edits, AI assistance usage, and acceptance rates.

```
POST /teams/daily-usage-data
```

#### Request Body

<Note>
  Date range cannot exceed 90 days. Make multiple requests for longer periods.
</Note>

#### Response

```typescript  theme={null}
{
  data: {
    date: number;
    isActive: boolean;
    totalLinesAdded: number;
    totalLinesDeleted: number;
    acceptedLinesAdded: number;
    acceptedLinesDeleted: number;
    totalApplies: number;
    totalAccepts: number;
    totalRejects: number;
    totalTabsShown: number;
    totalTabsAccepted: number;
    composerRequests: number;
    chatRequests: number;
    agentRequests: number;
    cmdkUsages: number;
    subscriptionIncludedReqs: number;
    apiKeyReqs: number;
    usageBasedReqs: number;
    bugbotUsages: number;
    mostUsedModel: string;
    applyMostUsedExtension?: string;
    tabMostUsedExtension?: string;
    clientVersion?: string;
    email?: string;
  }[];
  period: {
    startDate: number;
    endDate: number;
  };
}
```

#### Response Fields

#### Example Response

```json  theme={null}
{
  "data": [
    {
      "date": 1710720000000,
      "isActive": true,
      "totalLinesAdded": 1543,
      "totalLinesDeleted": 892,
      "acceptedLinesAdded": 1102,
      "acceptedLinesDeleted": 645,
      "totalApplies": 87,
      "totalAccepts": 73,
      "totalRejects": 14,
      "totalTabsShown": 342,
      "totalTabsAccepted": 289,
      "composerRequests": 45,
      "chatRequests": 128,
      "agentRequests": 12,
      "cmdkUsages": 67,
      "subscriptionIncludedReqs": 180,
      "apiKeyReqs": 0,
      "usageBasedReqs": 5,
      "bugbotUsages": 3,
      "mostUsedModel": "gpt-4",
      "applyMostUsedExtension": ".tsx",
      "tabMostUsedExtension": ".ts",
      "clientVersion": "0.25.1",
      "email": "developer@company.com"
    },
    {
      "date": 1710806400000,
      "isActive": true,
      "totalLinesAdded": 2104,
      "totalLinesDeleted": 1203,
      "acceptedLinesAdded": 1876,
      "acceptedLinesDeleted": 987,
      "totalApplies": 102,
      "totalAccepts": 91,
      "totalRejects": 11,
      "totalTabsShown": 456,
      "totalTabsAccepted": 398,
      "composerRequests": 67,
      "chatRequests": 156,
      "agentRequests": 23,
      "cmdkUsages": 89,
      "subscriptionIncludedReqs": 320,
      "apiKeyReqs": 15,
      "usageBasedReqs": 0,
      "bugbotUsages": 5,
      "mostUsedModel": "claude-3-opus",
      "applyMostUsedExtension": ".py",
      "tabMostUsedExtension": ".py",
      "clientVersion": "0.25.1",
      "email": "developer@company.com"
    }
  ],
  "period": {
    "startDate": 1710720000000,
    "endDate": 1710892800000
  }
}
```

#### Example Request

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/daily-usage-data \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "startDate": 1710720000000,
    "endDate": 1710892800000
  }'
```

### Get Spending Data

Retrieve spending information for the current calendar month with search, sorting, and pagination.

```
POST /teams/spend
```

#### Request Body

#### Response

```typescript  theme={null}
{
  teamMemberSpend: {
    spendCents: number;
    fastPremiumRequests: number;
    name: string;
    email: string;
    role: 'owner' | 'member' | 'free-owner';
    hardLimitOverrideDollars: number;
  }[];
  subscriptionCycleStart: number;
  totalMembers: number;
  totalPages: number;
}
```

#### Response Fields

#### Example Response

```json  theme={null}
{
  "teamMemberSpend": [
    {
      "spendCents": 2450,
      "fastPremiumRequests": 1250,
      "name": "Alex",
      "email": "developer@company.com",
      "role": "member",
      "hardLimitOverrideDollars": 100
    },
    {
      "spendCents": 1875,
      "fastPremiumRequests": 980,
      "name": "Sam",
      "email": "admin@company.com",
      "role": "owner",
      "hardLimitOverrideDollars": 0
    },
  ],
  "subscriptionCycleStart": 1708992000000,
  "totalMembers": 15,
  "totalPages": 1
}
```

#### Example Requests

**Basic spending data:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/spend \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{}'
```

**Search specific user with pagination:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/spend \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "searchTerm": "alex@company.com",
    "page": 2,
    "pageSize": 25
  }'
```

### Get Usage Events Data

Retrieve detailed usage events for your team with comprehensive filtering, search, and pagination options. This endpoint provides granular insights into individual API calls, model usage, token consumption, and costs.

```
POST /teams/filtered-usage-events
```

#### Request Body

#### Response

```typescript  theme={null}
{
  totalUsageEventsCount: number;
  pagination: {
    numPages: number;
    currentPage: number;
    pageSize: number;
    hasNextPage: boolean;
    hasPreviousPage: boolean;
  };
  usageEvents: {
    timestamp: string;
    model: string;
    kind: string;
    maxMode: boolean;
    requestsCosts: number;
    isTokenBasedCall: boolean;
    tokenUsage?: {
      inputTokens: number;
      outputTokens: number;
      cacheWriteTokens: number;
      cacheReadTokens: number;
      totalCents: number;
    };
    isFreeBugbot: boolean;
    userEmail: string;
  }[];
  period: {
    startDate: number;
    endDate: number;
  };
}
```

#### Response Fields Explained

#### Example Response

```json  theme={null}
{
  "totalUsageEventsCount": 113,
  "pagination": {
    "numPages": 12,
    "currentPage": 1,
    "pageSize": 10,
    "hasNextPage": true,
    "hasPreviousPage": false
  },
  "usageEvents": [
    {
      "timestamp": "1750979225854",
      "model": "claude-4-opus",
      "kind": "Usage-based",
      "maxMode": true,
      "requestsCosts": 5,
      "isTokenBasedCall": true,
      "tokenUsage": {
        "inputTokens": 126,
        "outputTokens": 450,
        "cacheWriteTokens": 6112,
        "cacheReadTokens": 11964,
        "totalCents": 20.18232
      },
      "isFreeBugbot": false,
      "userEmail": "developer@company.com"
    },
    {
      "timestamp": "1750979173824",
      "model": "claude-4-opus",
      "kind": "Usage-based",
      "maxMode": true,
      "requestsCosts": 10,
      "isTokenBasedCall": true,
      "tokenUsage": {
        "inputTokens": 5805,
        "outputTokens": 311,
        "cacheWriteTokens": 11964,
        "cacheReadTokens": 0,
        "totalCents": 40.16699999999999
      },
      "isFreeBugbot": false,
      "userEmail": "developer@company.com"
    },
    {
      "timestamp": "1750978339901",
      "model": "claude-4-sonnet-thinking",
      "kind": "Included in Business",
      "maxMode": true,
      "requestsCosts": 1.4,
      "isTokenBasedCall": false,
      "isFreeBugbot": false,
      "userEmail": "admin@company.com"
    }
  ],
  "period": {
    "startDate": 1748411762359,
    "endDate": 1751003762359
  }
}
```

#### Example Requests

**Get all usage events with default pagination:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/filtered-usage-events \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{}'
```

**Filter by date range and specific user:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/filtered-usage-events \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "startDate": 1748411762359,
    "endDate": 1751003762359,
    "email": "developer@company.com",
    "page": 1,
    "pageSize": 25
  }'
```

**Get usage events for a specific user with custom pagination:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/filtered-usage-events \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "userId": 12345,
    "page": 2,
    "pageSize": 50
  }'
```

### Set User Spend Limit

Set spending limits for individual team members. This allows you to control how much each user can spend on AI usage within your team.

```
POST /teams/user-spend-limit
```

<Note>
  **Rate limiting:** 60 requests per minute per team
</Note>

#### Request Body

<Note>
  * The user must already be a member of your team
  * Only integer values are accepted (no decimal amounts)
  * Setting `spendLimitDollars` to 0 will set the limit to \$0
</Note>

#### Response

Returns a standardized response indicating success or failure:

```typescript  theme={null}
{
  outcome: 'success' | 'error';
  message: string;
}
```

#### Example Responses

**Successfully set a limit:**

```json  theme={null}
{
  "outcome": "success",
  "message": "Spend limit set to $100 for user developer@company.com"
}
```

**Error response:**

```json  theme={null}
{
  "outcome": "error",
  "message": "Invalid email format"
}
```

#### Example Requests

**Set a spending limit:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/user-spend-limit \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "userEmail": "developer@company.com",
    "spendLimitDollars": 100
  }'
```

### Repo Blocklists API

Add repositories and use patterns to prevent files or directories from being indexed or used as context for your team.

#### Get Team Repo Blocklists

Retrieve all repository blocklists configured for your team.

```
GET /settings/repo-blocklists/repos
```

##### Response

Returns an array of repository blocklist objects:

```typescript  theme={null}
{
  repos: {
    id: string;
    url: string;
    patterns: string[];
  }[];
}
```

##### Example Response

```json  theme={null}
{
  "repos": [
    {
      "id": "repo_123",
      "url": "https://github.com/company/sensitive-repo",
      "patterns": ["*.env", "config/*", "secrets/**"]
    },
    {
      "id": "repo_456",
      "url": "https://github.com/company/internal-tools",
      "patterns": ["*"]
    }
  ]
}
```

##### Example Request

```bash  theme={null}
curl -X GET https://api.cursor.com/settings/repo-blocklists/repos \
  -u YOUR_API_KEY:
```

#### Upsert Repo Blocklists

Replace existing repository blocklists for the provided repos.
*Note: This endpoint will only overwrite the patterns for the repositories provided. All other repos will be unaffected.*

```
POST /settings/repo-blocklists/repos/upsert
```

##### Request Body

| Parameter | Type  | Required | Description                           |
| --------- | ----- | -------- | ------------------------------------- |
| repos     | array | Yes      | Array of repository blocklist objects |

Each repository object must contain:

| Field    | Type      | Required | Description                                               |
| -------- | --------- | -------- | --------------------------------------------------------- |
| url      | string    | Yes      | Repository URL to blocklist                               |
| patterns | string\[] | Yes      | Array of file patterns to block (glob patterns supported) |

##### Response

Returns the updated list of repository blocklists:

```typescript  theme={null}
{
  repos: {
    id: string;
    url: string;
    patterns: string[];
  }[];
}
```

##### Example Request

```bash  theme={null}
curl -X POST https://api.cursor.com/settings/repo-blocklists/repos/upsert \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "repos": [
      {
        "url": "https://github.com/company/sensitive-repo",
        "patterns": ["*.env", "config/*", "secrets/**"]
      },
      {
        "url": "https://github.com/company/internal-tools", 
        "patterns": ["*"]
      }
    ]
  }'
```

#### Delete Repo Blocklist

Remove a specific repository from the blocklist.

```
DELETE /settings/repo-blocklists/repos/:repoId
```

##### Parameters

| Parameter | Type   | Required | Description                              |
| --------- | ------ | -------- | ---------------------------------------- |
| repoId    | string | Yes      | ID of the repository blocklist to delete |

##### Response

Returns 204 No Content on successful deletion.

##### Example Request

```bash  theme={null}
curl -X DELETE https://api.cursor.com/settings/repo-blocklists/repos/repo_123 \
  -u YOUR_API_KEY:
```

#### Pattern Examples

Common blocklist patterns:

* `*` - Block entire repository
* `*.env` - Block all .env files
* `config/*` - Block all files in config directory
* `**/*.secret` - Block all .secret files in any subdirectory
* `src/api/keys.ts` - Block specific file

# AI Code Tracking API
Source: https://docs.cursor.com/en/account/teams/ai-code-tracking-api

Access AI-generated code analytics for your team's repositories

Access AI-generated code analytics for your team's repositories. This includes per-commit AI usage as well as granular accepted AI changes.

<Note>
  The API is in its first release. We're expanding capabilities based on feedback - let us know what endpoints you need!
</Note>

* **Availability**: Only for enterprise teams
* **Status**: Alpha (response shapes and fields may change)

## Authentication

All API requests require authentication using an API key. This API uses the same Admin API authentication as other endpoints.

For detailed authentication instructions, see [Admin API authentication](/en/account/teams/admin-api#authentication).

## Base URL

All API endpoints use:

```
https://api.cursor.com
```

## Rate Limits

* 5 requests per minute per team, per endpoint

## Query Parameters

All endpoints below accept the same query parameters via query string:

<Note>
  Responses return userId as an encoded external ID with the prefix user\_. This is stable for API consumption.
</Note>

## Semantics and How Metrics Are Computed

* **Sources**: "TAB" represents inline completions that were accepted; "COMPOSER" represents accepted diffs from Composer
* **Lines metrics**: tabLinesAdded/Deleted and composerLinesAdded/Deleted are separately counted; nonAiLinesAdded/Deleted are derived as max(0, totalLines - AI lines)
* **Privacy mode**: If enabled in the client, some metadata (like fileName) may be omitted
* **Branch info**: isPrimaryBranch is true when the current branch equals the repo's default branch; may be undefined if repo info is unavailable

You can scan that file to understand how commits and changes are detected and reported.

## Endpoints

### Get AI Commit Metrics (JSON, paginated)

Retrieve aggregated per-commit metrics that attribute lines to TAB, COMPOSER, and non-AI.

```
GET /analytics/ai-code/commits
```

#### Response

```typescript  theme={null}
{
  items: AiCommitMetric[];
  totalCount: number;
  page: number;
  pageSize: number;
}
```

#### AiCommitMetric Fields

#### Example Response

```json  theme={null}
{
  "items": [
    {
      "commitHash": "a1b2c3d4",
      "userId": "user_3k9x8q...",
      "userEmail": "developer@company.com",
      "repoName": "company/repo",
      "branchName": "main",
      "isPrimaryBranch": true,
      "totalLinesAdded": 120,
      "totalLinesDeleted": 30,
      "tabLinesAdded": 50,
      "tabLinesDeleted": 10,
      "composerLinesAdded": 40,
      "composerLinesDeleted": 5,
      "nonAiLinesAdded": 30,
      "nonAiLinesDeleted": 15,
      "message": "Refactor: extract analytics client",
      "commitTs": "2025-07-30T14:12:03.000Z",
      "createdAt": "2025-07-30T14:12:30.000Z"
    }
  ],
  "totalCount": 42,
  "page": 1,
  "pageSize": 100
}
```

#### Example Requests

**Basic request:**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/commits?startDate=7d&endDate=now&page=1&pageSize=100" \
  -u YOUR_API_KEY:
```

**Filter by user (email):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/commits?startDate=2025-06-01T00:00:00Z&endDate=now&user=developer@company.com" \
  -u YOUR_API_KEY:
```

### Download AI Commit Metrics (CSV, streaming)

Download commit metrics data in CSV format for large data extractions.

```
GET /analytics/ai-code/commits.csv
```

#### Response

Headers:

* Content-Type: text/csv; charset=utf-8

#### CSV Columns

#### Sample CSV Output

```csv  theme={null}
commit_hash,user_id,user_email,repo_name,branch_name,is_primary_branch,total_lines_added,total_lines_deleted,tab_lines_added,tab_lines_deleted,composer_lines_added,composer_lines_deleted,non_ai_lines_added,non_ai_lines_deleted,message,commit_ts,created_at
a1b2c3d4,user_3k9x8q...,developer@company.com,company/repo,main,true,120,30,50,10,40,5,30,15,"Refactor: extract analytics client",2025-07-30T14:12:03.000Z,2025-07-30T14:12:30.000Z
e5f6g7h8,user_3k9x8q...,developer@company.com,company/repo,feature-branch,false,85,15,30,5,25,3,30,7,"Add error handling",2025-07-30T13:45:21.000Z,2025-07-30T13:45:45.000Z
```

#### Example Request

```bash  theme={null}
curl -L "https://api.cursor.com/analytics/ai-code/commits.csv?startDate=2025-07-01T00:00:00Z&endDate=now&user=user_3k9x8q..." \
  -u YOUR_API_KEY: \
  -o commits.csv
```

### Get AI Code Change Metrics (JSON, paginated)

Retrieve granular accepted AI changes, grouped by deterministic changeId. Useful to analyze accepted AI events independent of commits.

```
GET /analytics/ai-code/changes
```

#### Response

```typescript  theme={null}
{
  items: AiCodeChangeMetric[];
  totalCount: number;
  page: number;
  pageSize: number;
}
```

#### AiCodeChangeMetric Fields

#### Example Response

```json  theme={null}
{
  "items": [
    {
      "changeId": "749356201",
      "userId": "user_3k9x8q...",
      "userEmail": "developer@company.com",
      "source": "COMPOSER",
      "model": null,
      "totalLinesAdded": 18,
      "totalLinesDeleted": 4,
      "createdAt": "2025-07-30T15:10:12.000Z",
      "metadata": [
        { "fileName": "src/analytics/report.ts", "fileExtension": "ts", "linesAdded": 12, "linesDeleted": 3 },
        { "fileName": "src/analytics/ui.tsx", "fileExtension": "tsx", "linesAdded": 6, "linesDeleted": 1 }
      ]
    }
  ],
  "totalCount": 128,
  "page": 1,
  "pageSize": 200
}
```

#### Example Requests

**Basic request:**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?startDate=14d&endDate=now&page=1&pageSize=200" \
  -u YOUR_API_KEY:
```

**Filter by user (encoded ID):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?user=user_3k9x8q..." \
  -u YOUR_API_KEY:
```

**Filter by user (email):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?user=developer@company.com" \
  -u YOUR_API_KEY:
```

### Download AI Code Change Metrics (CSV, streaming)

Download change metrics data in CSV format for large data extractions.

```
GET /analytics/ai-code/changes.csv
```

#### Response

Headers:

* Content-Type: text/csv; charset=utf-8

#### CSV Columns

#### Notes

* metadata\_json is a JSON stringified array of metadata entries (may omit fileName under privacy mode)
* When consuming CSV, be sure to parse quoted fields

#### Sample CSV Output

```csv  theme={null}
change_id,user_id,user_email,source,model,total_lines_added,total_lines_deleted,created_at,metadata_json
749356201,user_3k9x8q...,developer@company.com,COMPOSER,gpt-4o,18,4,2025-07-30T15:10:12.000Z,"[{""fileName"":""src/analytics/report.ts"",""fileExtension"":""ts"",""linesAdded"":12,""linesDeleted"":3},{""fileName"":""src/analytics/ui.tsx"",""fileExtension"":""tsx"",""linesAdded"":6,""linesDeleted"":1}]"
749356202,user_3k9x8q...,developer@company.com,TAB,,8,2,2025-07-30T15:08:45.000Z,"[{""fileName"":""src/utils/helpers.ts"",""fileExtension"":""ts"",""linesAdded"":8,""linesDeleted"":2}]"
```

#### Example Request

```bash  theme={null}
curl -L "https://api.cursor.com/analytics/ai-code/changes.csv?startDate=30d&endDate=now" \
  -u YOUR_API_KEY: \
  -o changes.csv
```

## Tips

* Use `user` parameter to quickly filter a single user across all endpoints
* For large data extractions, prefer CSV endpoints—they stream in pages of 10,000 records server-side
* `isPrimaryBranch` may be undefined if the client couldn't resolve the default branch
* `commitTs` is the commit timestamp; `createdAt` is the ingestion time on our servers
* Some fields may be absent when privacy mode is enabled on the client

## Changelog

* **Alpha release**: Initial endpoints for commits and changes. Response shapes may evolve based on feedback

# Analytics
Source: https://docs.cursor.com/en/account/teams/analytics

Track team usage and activity metrics

Team admins can track metrics from the [dashboard](/en/account/teams/dashboard).

### Total Usage

View aggregate metrics across your team, including total tabs and premium requests. For teams under 30 days old, metrics reflect usage since creation, including team members' pre-join activity.

### Per Active User

See average metrics per active user: tabs accepted, lines of code, and premium requests.

### User Activity

Track weekly and monthly active users.

## Analytics Report Headers

When you export analytics data from the dashboard, the report includes detailed metrics about user behavior and feature usage. Here's what each header means:

### User Information

<ResponseField name="Date" type="ISO 8601 timestamp">
  The date when the analytics data was recorded (e.g., 2024-01-15T04:30:00.000Z)
</ResponseField>

<ResponseField name="User ID" type="string">
  Unique identifier for each user in the system
</ResponseField>

<ResponseField name="Email" type="string">
  User's email address associated with their account
</ResponseField>

<ResponseField name="Is Active" type="boolean">
  Indicates if the user was active on this date
</ResponseField>

### AI-Generated Code Metrics

<ResponseField name="Chat Suggested Lines Added" type="number">
  Total lines of code suggested by the AI chat feature
</ResponseField>

<ResponseField name="Chat Suggested Lines Deleted" type="number">
  Total lines of code suggested for deletion by the AI chat
</ResponseField>

<ResponseField name="Chat Accepted Lines Added" type="number">
  AI-suggested lines that the user accepted and added to their code
</ResponseField>

<ResponseField name="Chat Accepted Lines Deleted" type="number">
  AI-suggested deletions that the user accepted
</ResponseField>

### Feature Usage Metrics

<ResponseField name="Chat Total Applies" type="number">
  Times a user applied AI-generated changes from chat
</ResponseField>

<ResponseField name="Chat Total Accepts" type="number">
  Times a user accepted AI suggestions
</ResponseField>

<ResponseField name="Chat Total Rejects" type="number">
  Times a user rejected AI suggestions
</ResponseField>

<ResponseField name="Chat Tabs Shown" type="number">
  Times AI suggestion tabs were displayed to the user
</ResponseField>

<ResponseField name="Tabs Accepted" type="number">
  AI suggestion tabs that were accepted by the user
</ResponseField>

### Request Type Metrics

<ResponseField name="Edit Requests" type="number">
  Requests made through the composer/edit feature (Cmd+K inline edits)
</ResponseField>

<ResponseField name="Ask Requests" type="number">
  Chat requests where users asked questions to the AI
</ResponseField>

<ResponseField name="Agent Requests" type="number">
  Requests made to AI agents (specialized AI assistants)
</ResponseField>

<ResponseField name="Cmd+K Usages" type="number">
  Times the Cmd+K (or Ctrl+K) command palette was used
</ResponseField>

### Subscription and API Metrics

<ResponseField name="Subscription Included Reqs" type="number">
  AI requests covered under the user's subscription plan
</ResponseField>

<ResponseField name="API Key Reqs" type="number">
  Requests made using API keys for programmatic access
</ResponseField>

<ResponseField name="Usage-Based Reqs" type="number">
  Requests that count toward usage-based billing
</ResponseField>

### Additional Features

<ResponseField name="Bugbot Usages" type="number">
  Times the bug detection/fixing AI feature was used
</ResponseField>

### Configuration Information

<ResponseField name="Most Used Model" type="string">
  The AI model that the user used most frequently (e.g., GPT-4, Claude)
</ResponseField>

<ResponseField name="Most Used Apply Extension" type="string">
  File extension most commonly used when applying AI suggestions (e.g., .ts,
  .py, .java)
</ResponseField>

<ResponseField name="Most Used Tab Extension" type="string">
  File extension most commonly used with tab completion features
</ResponseField>

<ResponseField name="Client Version" type="string">
  Version of the Cursor editor being used
</ResponseField>

### Calculated Metrics

The report also includes processed data that helps understand AI code contribution:

* **Total Lines Added/Deleted**: Raw count of all code changes
* **Accepted Lines Added/Deleted**: Lines that originated from AI suggestions and were accepted
* **Composer Requests**: Requests made through the inline composer feature
* **Chat Requests**: Requests made through the chat interface

<Note>
  All numeric values default to 0 if not present, boolean values default to
  false, and string values default to empty strings. Metrics are aggregated at
  the daily level per user.
</Note>

# Analytics V2
Source: https://docs.cursor.com/en/account/teams/analyticsV2

Advanced team usage and activity metrics tracking

We are working on a V2 release of our analytics infrastructure. This includes a refactor of the way in which we track various metrics.

As of **September 1, 2025**, and for users on **Cursor version 1.5**, analytics will utilize our V2 infrastructure. Prior versions would have undercounted various metrics, including:

* Total Accepted Lines of Code
* Total Suggested Lines of Code
* Total Tabs Accepted

Please stay tuned as we continue to invest in analytics and release new features in this space.

# Dashboard
Source: https://docs.cursor.com/en/account/teams/dashboard

Manage billing, usage, and team settings from your dashboard

The dashboard lets you access billing, set up usage-based pricing, and manage your Team.

## Overview

Get a quick summary of your team's activity, usage statistics, and recent changes. The overview page provides at-a-glance insights into your workspace.

## Settings

Configure team-wide preferences and security settings. The settings page includes:

## Teams & Enterprise Settings

<AccordionGroup>
  <Accordion title="Privacy Settings">
    Control data sharing preferences for your team. Configure zero data retention policies with AI providers (OpenAI, Anthropic, Google Vertex AI, xAi Grok) and manage team-wide privacy enforcement.
  </Accordion>

{" "}

<Accordion title="Usage-Based Pricing Settings">
    Enable usage-based pricing and set spending limits. Configure monthly team
    spending limits and optional per-user limits. Control whether only admins can
    modify these settings.
  </Accordion>

{" "}

<Accordion title="Bedrock IAM Role">
    Configure AWS Bedrock IAM roles for secure cloud integration.
  </Accordion>

{" "}

<Accordion title="Single Sign-On (SSO)">
    Set up SSO authentication for enterprise teams to streamline user access and
    improve security.
  </Accordion>

{" "}

<Accordion title="Cursor Admin API Keys">
    Create and manage API keys for programmatic access to Cursor's admin features.
  </Accordion>

{" "}

<Accordion title="Active Sessions">
    Monitor and manage active user sessions across your team.
  </Accordion>

<Accordion title="Invite Code Management">
    Create and manage invite codes for adding new team members.
  </Accordion>

<Accordion title="API Endpoints">
    Access Cursor's REST API endpoints for programmatic integration. All API endpoints are available on both Team and Enterprise plans, except for the [AI Code Tracking API](/en/docs/account/teams/ai-code-tracking-api) which requires Enterprise membership.
  </Accordion>
</AccordionGroup>

## Enterprise-Only Settings

<AccordionGroup>
  {" "}

<Accordion title="Model Access Control">
    Control which AI models are available to team members. Set restrictions on
    specific models or model tiers to manage costs and ensure appropriate usage
    across your organization.
  </Accordion>

{" "}

<Accordion title="Auto Run Configuration (0.49+)">
    Configure automatic command execution settings for Cursor version 0.49 and
    above. Control which commands can be executed automatically and set security
    policies for code execution.
  </Accordion>

<Accordion title="Repository Blocklist">
    Prevent access to specific repositories for security or compliance reasons.
  </Accordion>

{" "}

<Accordion title="MCP Configuration (0.51+)">
    Configure Model Context Protocol settings for Cursor version 0.51 and above.
    Manage how models access and process context from your development
    environment.
  </Accordion>

{" "}

<Accordion title="Cursor Ignore Configuration (0.50+)">
    Set up ignore patterns for files and directories in Cursor version 0.50 and
    above. Control which files and directories are excluded from AI analysis and
    suggestions.
  </Accordion>

<Accordion title=".cursor Directory Protection (0.51+)">
    Protect the .cursor directory from unauthorized access in version 0.51 and above. Ensure sensitive configuration and cache files remain secure.
  </Accordion>

<Accordion title="AI Code Tracking API">
    Access detailed AI-generated code analytics for your team's repositories. Retrieve per-commit AI usage metrics and granular accepted AI changes through REST API endpoints. Requires Enterprise plan. Find more information [here](/en/account/teams/ai-code-tracking-api).
  </Accordion>
</AccordionGroup>

<Note>
  **SCIM** (System for Cross-domain Identity Management) provisioning is also
  available for Enterprise plans. See our [SCIM
  documentation](/en/account/teams/scim) for setup instructions.
</Note>

## Members

Manage your team members, invite new users, and control access permissions. Set role-based permissions and monitor member activity.

## Integrations

{" "}

Connect Cursor with your favorite tools and services. Configure integrations with version control systems, project management tools, and other developer services.

## Background Agents

{" "}

Monitor and manage background agents running in your workspace. View agent status, logs, and resource usage.

## Bugbot

Access automated bug detection and fixing capabilities. Bugbot helps identify and resolve common issues in your codebase automatically.

## Active Directory Management

For enterprise teams, manage user authentication and access through Active Directory integration. Configure SSO and user provisioning.

## Usage

Track detailed usage metrics including AI requests, model usage, and resource consumption. Monitor usage across team members and projects.

## Billing & Invoices

Manage your subscription, update payment methods, and access billing history. Download invoices and manage usage-based pricing settings.

# Enterprise Settings
Source: https://docs.cursor.com/en/account/teams/enterprise-settings

Centrally manage Cursor settings for your organization

# Enterprise settings

You can centrally manage specific features of Cursor through device management solutions to ensure it meets the needs of your organization. When you specify a Cursor policy, its value overrides the corresponding Cursor setting on users' devices.

Settings editor showing that the 'Extensions: Allowed' setting is managed by the organization.

Cursor currently provides policies to control the following admin-controlled features:

| Policy            | Description                                                                                                | Cursor setting           | Available since |
| ----------------- | ---------------------------------------------------------------------------------------------------------- | ------------------------ | --------------- |
| AllowedExtensions | Controls which extensions can be installed.                                                                | extensions.allowed       | 1.2             |
| AllowedTeamId     | Controls which team IDs are allowed to log in. Users with unauthorized team IDs are forcefully logged out. | cursorAuth.allowedTeamId | 1.3             |

## Configure allowed extensions

The `extensions.allowed` Cursor setting controls which extensions can be installed. This setting accepts a JSON object where keys are publisher names and values are booleans indicating whether extensions from that publisher are allowed.

For example, setting `extensions.allowed` to `{"anysphere": true, "github": true}` allows extensions from Anysphere and GitHub publishers, while setting it to `{"anysphere": false}` blocks Anysphere extensions.

To centrally manage allowed extensions for your organization, configure the `AllowedExtensions` policy using your device management solution. This policy overrides the `extensions.allowed` setting on users' devices. The value of this policy is a JSON string that defines the allowed publishers.

If you want to learn more about extensions in Cursor, refer to the extensions documentation.

## Configure allowed team IDs

The `cursorAuth.allowedTeamId` Cursor setting controls which team IDs are permitted to log into Cursor. This setting accepts a comma-separated list of team IDs that are authorized for access.

For example, setting `cursorAuth.allowedTeamId` to `"1,3,7"` allows users from those specific team IDs to log in.

When a user attempts to log in with a team ID that is not in the allowed list:

* They are forcefully logged out immediately
* An error message is displayed
* The application prevents further authentication attempts until a valid team ID is used

To centrally manage allowed team IDs for your organization, configure the `AllowedTeamId` policy using your device management solution. This policy overrides the `cursorAuth.allowedTeamId` setting on users' devices. The value of this policy is a string containing the comma-separated list of authorized team IDs.

## Group Policy on Windows

Cursor has support for Windows Registry-based Group Policy. When policy definitions are installed, admins can use the Local Group Policy Editor to manage the policy values.

To add a policy:

1. Copy the Policy ADMX and ADML files from `AppData\Local\Programs\cursor\policies`.
2. Paste the ADMX file into the `C:\Windows\PolicyDefinitions` directory, and the ADML file into the `C:\Windows\PolicyDefinitions\<your-locale>\` directory.
3. Restart the Local Group Policy Editor.
4. Set the appropriate policy values (e.x. `{"anysphere": true, "github": true}` for the `AllowedExtensions` policy) in the Local Group Policy Editor.

Policies can be set both at the Computer level and the User level. If both are set, Computer level will take precedence. When a policy value is set, the value overrides the Cursor setting value configured at any level (default, user, workspace, etc.).

## Configuration profiles on macOS

Configuration profiles manage settings on macOS devices. A profile is an XML file with key/value pairs that correspond to available policy. These profiles can be deployed using Mobile Device Management (MDM) solutions, or installed manually.

<Accordion title="Example .mobileconfig file">
  An example `.mobileconfig` file for macOS is shown below:

```
  <?xml version="1.0" encoding="UTF-8"?>
  <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
  <plist version="1.0">
  	<dict>
  		<key>PayloadContent</key>
  		<array>
  			<dict>
  				<key>PayloadDisplayName</key>
  				<string>Cursor</string>
  				<key>PayloadIdentifier</key>
  				<string>com.todesktop.230313mzl4w4u92.J6B5723A-6539-4F31-8A4E-3CC96E51F48C</string>
  				<key>PayloadType</key>
  				<string>com.todesktop.230313mzl4w4u92</string>
  				<key>PayloadUUID</key>
  				<string>J6B5723A-6539-4F31-8A4E-3CC96E51F48C</string>
  				<key>PayloadVersion</key>
  				<integer>1</integer>
  				<key>AllowedExtensions</key>
  				<string>{"anysphere":true}</string>
  				<key>AllowedTeamId</key>
  				<string>1,2</string>
  			</dict>
  		</array>
  		<key>PayloadDescription</key>
  		<string>This profile manages Cursor.</string>
  		<key>PayloadDisplayName</key>
  		<string>Cursor</string>
  		<key>PayloadIdentifier</key>
  		<string>com.todesktop.230313mzl4w4u92</string>
  		<key>PayloadOrganization</key>
  		<string>Anysphere</string>
  		<key>PayloadType</key>
  		<string>Configuration</string>
  		<key>PayloadUUID</key>
  		<string>F2C1A7B3-9D4E-4B2C-8E1F-7A6C5D4B3E2F</string>
  		<key>PayloadVersion</key>
  		<integer>1</integer>
  		<key>TargetDeviceType</key>
  		<integer>5</integer>
  	</dict>
  </plist>
  ```
</Accordion>

### String policies

The example below demonstrates configuration of the `AllowedExtensions` policy. The policy value starts empty in the sample file (no extensions are allowed).

```
<key>AllowedExtensions</key>
<string></string>
```

Add the appropriate JSON string defining your policy between the `<string>` tags.

```
<key>AllowedExtensions</key>
<string>{"anysphere": true, "github": true}</string>
```

For the `AllowedTeamId` policy, add the comma-separated list of team IDs:

```
<key>AllowedTeamId</key>
<string>1,3,7</string>
```

**Important:** The provided `.mobileconfig` file initializes **all** policies available in that version of Cursor. Delete any policies that are not needed.

If you do not edit or remove a policy from the sample `.mobileconfig`, that policy will be enforced with its default (restrictive) policy value.

Manually install a configuration profile by double-clicking on the `.mobileconfig` profile in Finder and then enabling it in System Preferences under **General** > **Device Management**. Removing the profile from System Preferences will remove the policies from Cursor.

For more information on configuration profiles, refer to Apple's documentation.

## Additional policies

The goal is to promote current Cursor settings as policies and closely follow existing settings, so that the naming and behavior are consistent. If there are requests to enact more policies, please open an issue in the Cursor GitHub repository. The team will determine if there is already a corresponding setting for the behavior or if a new setting should be created to control the desired behavior.

## Frequently asked questions

### Does Cursor support configuration profiles on Linux?

Support for Linux is not on the roadmap. If you're interested in configuration profiles on Linux, open an issue in the Cursor GitHub repository and share details about your scenario.

# Members & Roles
Source: https://docs.cursor.com/en/account/teams/members

Manage team members and roles

Cursor teams have three roles:

## Roles

**Members** are the default role with access to Cursor's Pro features.

* Full access to Cursor's Pro features
* No access to billing settings or admin dashboard
* Can see their own usage and remaining usage-based budget

**Admins** control team management and security settings.

* Full access to Pro features
* Add/remove members, modify roles, setup SSO
* Configure usage-based pricing and spending limits
* Access to team analytics

**Unpaid Admins** manage teams without using a paid seat - ideal for IT or finance staff who don't need Cursor access.

* Not billable, no Pro features
* Same administrative capabilities as Admins

<Info>Unpaid Admins require at least one paid user on the team.</Info>

## Role Comparison

<div className="full-width-table">
  | Capability             | Member | Admin | Unpaid Admin |
  | ---------------------- | :----: | :---: | :----------: |
  | Use Cursor features    |    ✓   |   ✓   |              |
  | Invite members         |    ✓   |   ✓   |       ✓      |
  | Remove members         |        |   ✓   |       ✓      |
  | Change user role       |        |   ✓   |       ✓      |
  | Admin dashboard        |        |   ✓   |       ✓      |
  | Configure SSO/Security |        |   ✓   |       ✓      |
  | Manage Billing         |        |   ✓   |       ✓      |
  | View Analytics         |        |   ✓   |       ✓      |
  | Manage Access          |        |   ✓   |       ✓      |
  | Set usage controls     |        |   ✓   |       ✓      |
  | Requires paid seat     |    ✓   |   ✓   |              |
</div>

## Managing members

All team members can invite others. We don't currently control invites.

### Add member

Add members three ways:

1. **Email invitation**

* Click `Invite Members`
   * Enter email addresses
   * Users receive email invites

2. **Invite link**

* Click `Invite Members`
   * Copy `Invite Link`
   * Share with team members

3. **SSO**
   * Configure SSO in [admin dashboard](/en/account/teams/sso)
   * Users auto-join when logging in via SSO email

<Warning>
  Invite links have a long expiration date - anyone with the link can join.
  Revoke them or use [SSO](/en/account/teams/sso)
</Warning>

### Remove member

Admins can remove members anytime via context menu → "Remove". If a member has used any credits, their seat remains occupied until the end of the billing cycle.

### Change role

Admins can change roles for other members by clicking the context menu and then use the "Change role" option.<br />

There must be at least one Admin, and one paid member on the team at all times.

## Security & SSO

SAML 2.0 Single Sign-On (SSO) is available on Team plans. Key features include:

* Configure SSO connections ([learn more](/en/account/teams/sso))
* Set up domain verification
* Automatic user enrollment
* SSO enforcement options
* Identity provider integration (Okta, etc)

<Note>
  <p className="!mb-0">Domain verification is required to enable SSO.</p>
</Note>

## Usage Controls

Access usage settings to:

* Enable usage-based pricing
* Enable for premium models
* Set admin-only modifications
* Set monthly spending limits
* Monitor team-wide usage

## Billing

When adding team members:

* Each member or admin adds a billable seat (see [pricing](https://cursor.com/pricing))
* New members are charged pro-rata for their remaining time in the billing period
* Unpaid admin seats aren't counted

Mid-month additions charge only for days used. When removing members who have used credits, their seat remains occupied until the end of the billing cycle - no pro-rated refunds are given.

Role changes (e.g., Admin to Unpaid Admin) adjust billing from the change date. Choose monthly or yearly billing.

Monthly/yearly renewal occurs on your original signup date, regardless of member changes.

### Switch to Yearly billing

Save **20%** by switching from monthly to yearly:

1. Go to [Dashboard](https://cursor.com/dashboard)
2. In account section, click "Advanced" then "Upgrade to yearly billing"

<Note>
  You can only switch from monthly to yearly via dashboard. To switch from
  yearly to monthly, contact [hi@cursor.com](mailto:hi@cursor.com).
</Note>

# SCIM
Source: https://docs.cursor.com/en/account/teams/scim

Set up SCIM provisioning for automated user and group management

## Overview

SCIM 2.0 provisioning automatically manages your team members and directory groups through your identity provider. Available on Enterprise plans with SSO enabled.

<product_visual type="screenshot">
  SCIM settings dashboard showing Active Directory Management configuration
</product_visual>

## Prerequisites

* Cursor Enterprise plan
* SSO must be configured first - **SCIM requires an active SSO connection**
* Admin access to your identity provider (Okta, Azure AD, etc.)
* Admin access to your Cursor organization

## How it works

### User provisioning

Users are automatically added to Cursor when assigned to the SCIM application in your identity provider. When unassigned, they're removed. Changes sync in real-time.

### Directory groups

Directory groups and their membership sync from your identity provider. Group and user management must be done through your identity provider - Cursor displays this information as read-only.

### Spend management

Set different per-user spend limits for each directory group. Directory group limits take precedence over team-level limits. Users in multiple groups receive the highest applicable spend limit.

## Setup

<Steps>
  <Step title="Ensure SSO is configured">
    SCIM requires SSO to be set up first. If you haven't configured SSO yet,
    follow the [SSO setup guide](/en/account/teams/sso) before proceeding.
  </Step>

<Step title="Access Active Directory Management">
    Navigate to
    [cursor.com/dashboard?tab=active-directory](https://www.cursor.com/dashboard?tab=active-directory)
    with an admin account, or go to your dashboard settings and select the
    "Active Directory Management" tab.
  </Step>

<Step title="Start SCIM setup">
    Once SSO is verified, you'll see a link for step-by-step SCIM setup. Click
    this to begin the configuration wizard.
  </Step>

<Step title="Configure SCIM in your identity provider">
    In your identity provider: - Create or configure your SCIM application - Use
    the SCIM endpoint and token provided by Cursor - Enable user and push group
    provisioning - Test the connection
  </Step>

<Step title="Configure spend limits (optional)">
    Back in Cursor's Active Directory Management page: - View your synchronized
    directory groups - Set per-user spend limits for specific groups as needed -
    Review which limits apply to users in multiple groups
  </Step>
</Steps>

### Identity provider setup

For provider-specific setup instructions:

<Card title="Identity Provider Guides" icon="book" href="https://workos.com/docs/integrations">
  Setup instructions for Okta, Azure AD, Google Workspace, and more.
</Card>

## Managing users and groups

<Warning>
  All user and group management must be done through your identity provider.
  Changes made in your identity provider will automatically sync to Cursor, but
  you cannot modify users or groups directly in Cursor.
</Warning>

### User management

* Add users by assigning them to your SCIM application in your identity provider
* Remove users by unassigning them from the SCIM application
* User profile changes (name, email) sync automatically from your identity provider

### Group management

* Directory groups are automatically synced from your identity provider
* Group membership changes are reflected in real-time
* Use groups to organize users and set different spend limits

### Spend limits

* Set different per-user limits for each directory group
* Users inherit the highest spend limit from their groups
* Group limits override the default team-wide per-user limit

## FAQ

### Why isn't SCIM management showing up in my dashboard?

Ensure SSO is properly configured and working before setting up SCIM. SCIM requires an active SSO connection to function.

### Why aren't users syncing?

Verify that users are assigned to the SCIM application in your identity provider. Users must be explicitly assigned to appear in Cursor.

### Why aren't groups appearing?

Check that push group provisioning is enabled in your identity provider's SCIM settings. Group sync must be configured separately from user sync.

### Why aren't spend limits applying?

Confirm users are properly assigned to the expected groups in your identity provider. Group membership determines which spend limits apply.

### Can I manage SCIM users and groups directly in Cursor?

No. All user and group management must be done through your identity provider. Cursor displays this information as read-only.

### How quickly do changes sync?

Changes made in your identity provider sync to Cursor in real-time. There may be a brief delay for large bulk operations.

# Get Started
Source: https://docs.cursor.com/en/account/teams/setup

Create and set up a Cursor team

## Cursor for Teams

Cursor works for individuals and teams. The Teams plan provides tools for organizations: SSO, team management, access controls, and usage analytics.

## Creating a Team

Create a team by following these steps:

<Steps>
  <Step title="Set up Teams plan">
    To create a Team, follow these steps:

1. **For new users**: Visit [cursor.com/team/new-team](https://cursor.com/team/new-team) to create a new account and team
    2. **For existing users**: Go to your [dashboard](/en/account/dashboard) and click "Upgrade to Teams"
  </Step>

<Step title="Enter Team details">
    Select a Team name and billing cycle

<Step title="Invite members">
    Invite team members. User counts are prorated - you only pay for the time users are members.

<Step title="Enable SSO (optional)">
    Enable [SSO](/en/account/teams/sso) for security and automated onboarding.

## FAQ

<AccordionGroup>
  <Accordion title="My team uses ZScalar / a proxy / a VPN, will Cursor work?">
    Cursor uses HTTP/2 by default. Some proxies and VPNs block this.

Enable HTTP/1.1 fallback in settings to use HTTP/1.1 instead.
  </Accordion>

<Accordion title="How can I purchase licenses for my company?">
    Cursor bills per active user, not seats. Add or remove users anytime - new members are charged pro-rata for their remaining time. If a removed user has used any credits, their seat remains occupied until the end of the billing cycle.

Your renewal date stays the same.
  </Accordion>

<Accordion title="How can I set up a team when I'm not using Cursor?">
    Set yourself as an [Unpaid Admin](/en/account/teams/members) to manage without a license.

<Warning>
      Teams need at least one paid member. You can set up, invite a member, then change your role before billing.
    </Warning>
  </Accordion>

<Accordion title="How can I add Cursor to my company's MDM?">
    Download links for all platforms are available at [cursor.com/downloads](https://cursor.com/downloads).

MDM instructions:

* [Omnissa Workspace ONE](https://docs.omnissa.com/bundle/MobileApplicationManagementVSaaS/page/DeployInternalApplications.html) (formerly VMware)
    * [Microsoft Intune (Windows)](https://learn.microsoft.com/en-us/mem/intune-service/apps/apps-win32-app-management)
    * [Microsoft Intune (Mac)](https://learn.microsoft.com/en-us/mem/intune-service/apps/lob-apps-macos-dmg)
    * [Kandji MDM](https://support.kandji.io/kb/custom-apps-overview)
  </Accordion>
</AccordionGroup>

# SSO
Source: https://docs.cursor.com/en/account/teams/sso

Set up single sign-on for your team

## Overview

SAML 2.0 SSO is available at no additional cost on Business plans. Use your existing identity provider (IdP) to authenticate team members without separate Cursor accounts.

## Prerequisites

* Cursor Team plan
* Admin access to your identity provider (e.g., Okta)
* Admin access to your Cursor organization

## Configuration Steps

<Steps>
  <Step title="Sign in to your Cursor account">
    Navigate to [cursor.com/dashboard?tab=settings](https://www.cursor.com/dashboard?tab=settings) with an admin account.
  </Step>

<Step title="Locate the SSO configuration">
    Find the "Single Sign-On (SSO)" section and expand it.
  </Step>

<Step title="Begin the setup process">
    Click the "SSO Provider Connection settings"button to start SSO setup and follow the wizard.
  </Step>

<Step title="Configure your identity provider">
    In your identity provider (e.g., Okta):

* Create new SAML application
    * Configure SAML settings using Cursor's information
    * Set up Just-in-Time (JIT) provisioning
  </Step>

<Step title="Verify domain">
    Verify the domain of your users in Cursor by clicking the "Domain verification settings" button.
  </Step>
</Steps>

### Identity Provider Setup Guides

For provider-specific setup instructions:

<Card title="Identity Provider Guides" icon="book" href="https://workos.com/docs/integrations">
  Setup instructions for Okta, Azure AD, Google Workspace, and more.
</Card>

## Additional Settings

* Manage SSO enforcement through admin dashboard
* New users auto-enroll when signing in through SSO
* Handle user management through your identity provider

## Troubleshooting

If issues occur:

* Verify domain is verified in Cursor
* Ensure SAML attributes are properly mapped
* Check SSO is enabled in admin dashboard
* Match first and last names between identity provider and Cursor
* Check provider-specific guides above
* Contact [hi@cursor.com](mailto:hi@cursor.com) if issues persist

# Update Access
Source: https://docs.cursor.com/en/account/update-access

Choose how often you receive updates

export const Kbd = ({children, tooltip, os}) => {
  const keysInput = typeof children === 'string' && children.trim() !== '' ? children : null;
  if (!keysInput) {
    return null;
  }
  const isModifier = key => {
    const modifiers = ['⌘', '⇧', '⌥', '⌃', '⏎', '⌫', '⌦', '⎋', '⇥', '⌁', '←', '→', '↑', '↓', 'Ctrl', 'Shift', 'Alt', 'Cmd', 'Opt', 'Return', 'Backspace', 'Delete', 'Escape', 'Tab', 'Space', 'Enter', 'Esc', 'ArrowLeft', 'ArrowRight', 'ArrowUp', 'ArrowDown', 'Left', 'Right', 'Up', 'Down'];
    return modifiers.includes(key.trim());
  };
  const capitalizeFirstLetter = string => {
    return string.charAt(0).toUpperCase() + string.slice(1);
  };
  const isMac = os ? os.toLowerCase() === 'mac' || os.toLowerCase() === 'macos' : typeof navigator !== 'undefined' && (navigator.platform.toUpperCase().indexOf('MAC') >= 0 || navigator.userAgent.toUpperCase().indexOf('MAC') >= 0);
  const convertToSymbols = shortcut => {
    if (isMac) {
      return shortcut.replace(/⌘|Cmd|CMD/gi, '⌘').replace(/⌥|Opt|OPT/gi, '⌥').replace(/⌃|Ctrl/gi, '⌃').replace(/⇧|Shift/gi, '⇧').replace(/⏎|Return/gi, '⏎').replace(/⌫|Backspace/gi, '⌫').replace(/⌦|Delete/gi, '⌦').replace(/␛|Escape/gi, '␛').replace(/⇥|Tab/gi, '⇥').replace(/⌁|Space/gi, '⌁').replace(/←|Arrow\s*Left|ArrowLeft|Left/gi, '←').replace(/→|Arrow\s*Right|ArrowRight|Right/gi, '→').replace(/↑|Arrow\s*Up|ArrowUp|Up/gi, '↑').replace(/↓|Arrow\s*Down|ArrowDown|Down/gi, '↓');
    } else {
      const converted = shortcut.replace(/⌘|Cmd|CMD/gi, 'Ctrl').replace(/⌥|Opt|OPT/gi, 'Alt').replace(/⌃|Ctrl/gi, 'Ctrl').replace(/⇧|Shift/gi, 'Shift').replace(/⏎|Return/gi, 'Enter').replace(/⌫|Backspace/gi, 'Backspace').replace(/⌦|Delete/gi, 'Delete').replace(/⎋|Escape/gi, 'Esc').replace(/⇥|Tab/gi, 'Tab').replace(/⌁|Space/gi, 'Space').replace(/←|Arrow\s*Left|ArrowLeft|Left/gi, 'Arrow-Left').replace(/→|Arrow\s*Right|ArrowRight|Right/gi, 'Arrow-Right').replace(/↑|Arrow\s*Up|ArrowUp|Up/gi, 'Arrow-Up').replace(/↓|Arrow\s*Down|ArrowDown|Down/gi, 'Arrow-Down');
      const keyList = converted.split(/[\+\s]+/).filter(key => key.trim());
      return keyList.join('+');
    }
  };
  const convertToReadableText = shortcut => {
    const converted = shortcut.replace(/⌘|Cmd|CMD/gi, 'Cmd').replace(/⌥|Opt|OPT/gi, 'Opt').replace(/⌃|Ctrl/gi, 'Ctrl').replace(/⇧|Shift/gi, 'Shift').replace(/⏎|Return/gi, 'Return').replace(/⌫|Backspace/gi, 'Backspace').replace(/⌦|Delete/gi, 'Delete').replace(/⎋|Escape/gi, 'Escape').replace(/⇥|Tab/gi, 'Tab').replace(/⌁|Space/gi, 'Space').replace(/←|Arrow\s*Left|ArrowLeft|Left/gi, 'Arrow-Left').replace(/→|Arrow\s*Right|ArrowRight|Right/gi, 'Arrow-Right').replace(/↑|Arrow\s*Up|ArrowUp|Up/gi, 'Arrow-Up').replace(/↓|Arrow\s*Down|ArrowDown|Down/gi, 'Arrow-Down');
    const keyList = converted.split(/[\+\s]+/).filter(key => key.trim());
    return keyList.map(key => {
      const trimmedKey = key.trim();
      return isModifier(trimmedKey) ? trimmedKey : capitalizeFirstLetter(trimmedKey);
    }).join('+');
  };
  const displayShortcut = convertToSymbols(keysInput);
  const tooltipText = isMac ? tooltip ? `${convertToReadableText(keysInput)}: ${tooltip}` : convertToReadableText(keysInput) : tooltip || null;
  const processedKeys = isMac ? displayShortcut.split(/[\+\s]+/).filter(key => key.trim()).map(key => {
    const trimmedKey = key.trim();
    return isModifier(trimmedKey) ? trimmedKey : capitalizeFirstLetter(trimmedKey);
  }).join('') : displayShortcut.split('+').map(key => {
    const trimmedKey = key.trim();
    return isModifier(trimmedKey) ? trimmedKey : capitalizeFirstLetter(trimmedKey);
  }).join('+');
  return tooltipText ? <Tooltip tip={tooltipText}>
      <kbd>
        {processedKeys}
      </kbd>
    </Tooltip> : <kbd>
      {processedKeys}
    </kbd>;
};

Cursor has two update channels.

<Tabs>
  <Tab title="Default">
    The default update channel with tested releases.

* Stable releases
    * Bug fixes from pre-release testing
    * Default for all users
    * Only option for team users

<Note>
      Team and business accounts use Default mode.
    </Note>
  </Tab>

<Tab title="Early Access">
    Pre-release versions with new features.

<Warning>
      Early Access builds may have bugs or stability issues.
    </Warning>

* Access to features in development
    * May contain bugs
    * Not available for team accounts
  </Tab>
</Tabs>

## Change update channel

1. **Open settings**: Press <Kbd>Cmd+Shift+J</Kbd>
2. **Go to Beta**: Select Beta in the sidebar
3. **Select channel**: Choose Default or Early Access

Report Early Access issues on the [Forum](https://forum.cursor.com).

# Apply
Source: https://docs.cursor.com/en/agent/apply

Learn how to apply, accept, or reject code suggestions from chat using Apply

## How Apply works

Apply is a specialized Cursor model that takes code generated by chat and integrates it into your files. It processes the code blocks from chat conversations and applies the changes to your codebase.

Apply does not generate code itself. The chat model generates the code, and Apply handles the integration into existing files. It can process changes across multiple files and large codebases.

## Apply code blocks

To apply a code block suggestion, press the play button in the top right corner of a code block.

# Checkpoints
Source: https://docs.cursor.com/en/agent/chat/checkpoints

Save and restore previous states after Agent changes

Checkpoints are automatic snapshots of Agent's changes to your codebase. They let you undo Agent modifications if needed.

## Restoring checkpoints

Two ways to restore:

1. **From input box**: Click `Restore Checkpoint` button on previous requests
2. **From message**: Click the + button when hovering over a message

<Warning>
  Checkpoints are not version control. Use Git for permanent history.
</Warning>

## How they work

* Stored locally, separate from Git
* Track only Agent changes (not manual edits)
* Cleaned up automatically

<Note>
  Manual edits aren't tracked. Only use checkpoints for Agent changes.
</Note>

## FAQ

<AccordionGroup>
  <Accordion title="Do checkpoints affect Git?">
    No. They're separate from Git history.
  </Accordion>

{" "}

<Accordion title="How long are they kept?">
    For the current session and recent history. Automatically cleaned up.
  </Accordion>

<Accordion title="Can I create them manually?">
    No. They're created automatically by Cursor.
  </Accordion>
</AccordionGroup>

{" "}

# Commands
Source: https://docs.cursor.com/en/agent/chat/commands

Define commands for reusable workflows

Custom commands allow you to create reusable workflows that can be triggered with a simple `/` prefix in the chat input box. These commands help standardize processes across your team and make common tasks more efficient.

<Info>
  Commands are currently in beta. The feature and syntax may change as we continue to improve it.
</Info>

## How commands work

Commands are defined as plain Markdown files that can be stored in two locations:

1. **Project commands**: Stored in the `.cursor/commands` directory of your project
2. **Global commands**: Stored in the `~/.cursor/commands` directory in your home directory

When you type `/` in the chat input box, Cursor will automatically detect and display available commands from both directories, making them instantly accessible across your workflow.

## Creating commands

1. Create a `.cursor/commands` directory in your project root
2. Add `.md` files with descriptive names (e.g., `review-code.md`, `write-tests.md`)
3. Write plain Markdown content describing what the command should do
4. Commands will automatically appear in the chat when you type `/`

Here's an example of how your commands directory structure might look:

```
.cursor/
└── commands/
    ├── address-github-pr-comments.md
    ├── code-review-checklist.md
    ├── create-pr.md
    ├── light-review-existing-diffs.md
    ├── onboard-new-developer.md
    ├── run-all-tests-and-fix.md
    ├── security-audit.md
    └── setup-new-feature.md
```

## Examples

Try these commands in your projects to get a feel for how they work.

<AccordionGroup>
  <Accordion title="Code review checklist">
    ```markdown  theme={null}
    # Code Review Checklist

## Overview
    Comprehensive checklist for conducting thorough code reviews to ensure quality, security, and maintainability.

## Review Categories

### Functionality
    - [ ] Code does what it's supposed to do
    - [ ] Edge cases are handled
    - [ ] Error handling is appropriate
    - [ ] No obvious bugs or logic errors

### Code Quality
    - [ ] Code is readable and well-structured
    - [ ] Functions are small and focused
    - [ ] Variable names are descriptive
    - [ ] No code duplication
    - [ ] Follows project conventions

### Security
    - [ ] No obvious security vulnerabilities
    - [ ] Input validation is present
    - [ ] Sensitive data is handled properly
    - [ ] No hardcoded secrets
    ```
  </Accordion>

<Accordion title="Security audit">
    ```markdown  theme={null}
    # Security Audit

## Overview
    Comprehensive security review to identify and fix vulnerabilities in the codebase.

## Steps
    1. **Dependency audit**
       - Check for known vulnerabilities
       - Update outdated packages
       - Review third-party dependencies

2. **Code security review**
       - Check for common vulnerabilities
       - Review authentication/authorization
       - Audit data handling practices

3. **Infrastructure security**
       - Review environment variables
       - Check access controls
       - Audit network security

## Security Checklist
    - [ ] Dependencies updated and secure
    - [ ] No hardcoded secrets
    - [ ] Input validation implemented
    - [ ] Authentication secure
    - [ ] Authorization properly configured
    ```
  </Accordion>

<Accordion title="Setup new feature">
    ```markdown  theme={null}
    # Setup New Feature

## Overview
    Systematically set up a new feature from initial planning through to implementation structure.

## Steps
    1. **Define requirements**
       - Clarify feature scope and goals
       - Identify user stories and acceptance criteria
       - Plan technical approach

2. **Create feature branch**
       - Branch from main/develop
       - Set up local development environment
       - Configure any new dependencies

3. **Plan architecture**
       - Design data models and APIs
       - Plan UI components and flow
       - Consider testing strategy

## Feature Setup Checklist
    - [ ] Requirements documented
    - [ ] User stories written
    - [ ] Technical approach planned
    - [ ] Feature branch created
    - [ ] Development environment ready
    ```
  </Accordion>

<Accordion title="Create pull request">
    ```markdown  theme={null}
    # Create PR

## Overview
    Create a well-structured pull request with proper description, labels, and reviewers.

## Steps
    1. **Prepare branch**
       - Ensure all changes are committed
       - Push branch to remote
       - Verify branch is up to date with main

2. **Write PR description**
       - Summarize changes clearly
       - Include context and motivation
       - List any breaking changes
       - Add screenshots if UI changes

3. **Set up PR**
       - Create PR with descriptive title
       - Add appropriate labels
       - Assign reviewers
       - Link related issues

## PR Template
    - [ ] Feature A
    - [ ] Bug fix B
    - [ ] Unit tests pass
    - [ ] Manual testing completed
    ```
  </Accordion>

<Accordion title="Run tests and fix failures">
    ```markdown  theme={null}
    # Run All Tests and Fix Failures

## Overview
    Execute the full test suite and systematically fix any failures, ensuring code quality and functionality.

## Steps
    1. **Run test suite**
       - Execute all tests in the project
       - Capture output and identify failures
       - Check both unit and integration tests

2. **Analyze failures**
       - Categorize by type: flaky, broken, new failures
       - Prioritize fixes based on impact
       - Check if failures are related to recent changes

3. **Fix issues systematically**
       - Start with the most critical failures
       - Fix one issue at a time
       - Re-run tests after each fix
    ```
  </Accordion>

<Accordion title="Onboard new developer">
    ```markdown  theme={null}
    # Onboard New Developer

## Overview
    Comprehensive onboarding process to get a new developer up and running quickly.

## Steps
    1. **Environment setup**
       - Install required tools
       - Set up development environment
       - Configure IDE and extensions
       - Set up git and SSH keys

2. **Project familiarization**
       - Review project structure
       - Understand architecture
       - Read key documentation
       - Set up local database

## Onboarding Checklist
    - [ ] Development environment ready
    - [ ] All tests passing
    - [ ] Can run application locally
    - [ ] Database set up and working
    - [ ] First PR submitted
    ```
  </Accordion>
</AccordionGroup>

# Compact
Source: https://docs.cursor.com/en/agent/chat/compact

Save space in chat with compact mode interface

Compact mode provides a streamlined chat interface by reducing visual clutter and maximizing available space for conversations.

## Overview

When enabled, compact mode transforms the chat interface by:

* **Hiding icons** for a cleaner, minimalist appearance
* **Auto-collapsing diffs** to reduce visual noise
* **Auto-collapsing input** to maximize conversation space

This setting is particularly useful when working on smaller screens or when you prefer a focused, distraction-free chat experience.

## Before and After

### Default mode

### Compact mode

## Enabling compact mode

To enable compact mode:

1. Open Cursor Settings
2. Navigate to **Chat** settings
3. Toggle **Compact Mode** to enable

The interface will immediately update to the streamlined view, giving you more space to focus on your conversations.

# Duplicate
Source: https://docs.cursor.com/en/agent/chat/duplicate

Create branches from any point in a conversation

Duplicate/fork chats to explore alternative solutions without losing your current conversation.

## How to duplicate

1. Find where you want to branch off
2. Click the three dots on the message
3. Select "Duplicate Chat"

## What happens

* Context up to that point is preserved
* Original conversation remains unchanged
* Both chats maintain separate history

# Export
Source: https://docs.cursor.com/en/agent/chat/export

Export chats to markdown format

Export Agent chats as markdown files for sharing or documentation.

## What's exported

* All messages and responses
* Code blocks with syntax highlighting
* File references and context
* Chronological conversation flow

## How to export

1. Navigate to the chat to export
2. Click context menu → "Export Chat"
3. Save the file locally

<Warning>
  Review exports for sensitive data: API keys, internal URLs, proprietary code,
  personal information
</Warning>

# History
Source: https://docs.cursor.com/en/agent/chat/history

View and manage chat conversations

Access past Agent conversations from the history panel.

## Opening history

* Click the history icon in Agent sidepane
* Press <Kbd tooltip="Open chat history">Opt Cmd '</Kbd>

## Managing chats

* **Edit titles**: Click to rename
* **Delete**: Remove unneeded chats
* **Open**: Click to review full conversation

Chat history is stored locally in a SQLite database on your machine.

<Note>
  To preserve chats, [export them](/en/agent/chats/export) as markdown.
</Note>

## Background Agents

Background agent chats aren't in regular history but instead stored in a remote database. Use <Kbd tooltip="Open background agent control panel">Cmd E</Kbd> to view them.

## Referencing past chats

Use [@Past Chats](/en/context/@-symbols/@-past-chats) to include context from previous conversations in your current chat.

# Summarization
Source: https://docs.cursor.com/en/agent/chat/summarization

Context management for long conversations in chat

## Message summarization

As conversations grow longer, Cursor automatically summarizes and manages context to keep your chats efficient. Learn how to use the context menu and understand how files are condensed to fit within model context windows.

### Using the /summarize command

You can manually trigger summarization using the `/summarize` command in chat. This command helps manage context when conversations become too long, allowing you to continue working efficiently without losing important information.

<Info>
  For a deeper dive into how context works in Cursor, check our [Working with
  Context](/en/guides/working-with-context) guide.
</Info>

### How summarization works

When conversations grow longer, they exceed the model's context window limit:

<div className="bg-white px-3 py-2 mb-2 rounded border border-dashed border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="relative my-4">
        <div className="absolute top-0 left-1/2 transform -translate-x-1/2 -translate-y-1/2 text-center text-xs text-black bg-neutral-100 px-2">Context window limit</div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>
    </div>
  </div>
</Frame>

To solve this, Cursor summarizes older messages to make room for new conversations.

<Frame>
  <div className="font-mono text-xs w-full bg-neutral-100 p-4 rounded-lg">
    <div className="relative">
      <div className="relative my-4">
        <div className="absolute top-0 left-1/2 transform -translate-x-1/2 -translate-y-1/2 text-center text-xs text-black bg-neutral-100 px-2">
          Context window limit
        </div>

<div className="bg-neutral-100 px-3 py-2 mb-2 rounded border border-dashed border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">
          Summarized Messages
        </div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>
    </div>
  </div>
</Frame>

## File & folder condensation

While chat summarization handles long conversations, Cursor uses a different strategy for managing large files and folders: **smart condensation**. When you include files in your conversation, Cursor determines the best way to present them based on their size and available context space.

Here are the different states a file/folder can be in:

### Condensed

When files or folders are too large to fit within the context window, Cursor automatically condenses them. Condensing shows the model key structural elements like function signatures, classes, and methods. From this condensed view, the model can choose to expand specific files if needed. This approach maximizes effective use of the available context window.

### Significantly condensed

When a file name appears with a "Significantly Condensed" label, the file was too large to include in full, even in condensed form. Only the file name will be shown to the model.

### Not included

When a warning icon appears next to a file or folder, the item is too large to be included in the context window, even in condensed form. This helps you understand which parts of your codebase are accessible to the model.

# Tabs
Source: https://docs.cursor.com/en/agent/chat/tabs

Run multiple Agent conversations simultaneously

## Overview

Press <Kbd>Cmd+T</Kbd> to create new tabs. Each tab maintains separate conversation history, context, and model selection.

<Tip>
  For parallel workflows, try [Background Agents](/en/background-agents)
</Tip>

## Managing tabs

* Create new tabs with <Kbd>Cmd+T</Kbd>. Each tab starts with a fresh conversation and maintains its own context.

* Switch between tabs by clicking their headers or using <Kbd>Ctrl+Tab</Kbd> to cycle through them.

* Tab titles are auto-generated after the first message, but you can rename them by right-clicking on the tab header.

<Tip>
  Use one task per tab, provide clear initial descriptions, and close finished
  tabs to keep your workspace organized.
</Tip>

### Conflicts

Cursor prevents multiple tabs from editing the same files. You'll be prompted to resolve conflicts.

## Reference other chats

Use [@Past Chats](/en/context/@-symbols/@-past-chats) to include context from other tabs or previous sessions.

# Modes
Source: https://docs.cursor.com/en/agent/modes

Choose the right mode for your task - from autonomous coding to focused edits

Agent offers different modes optimized for specific tasks. Each mode has different capabilities and tools enabled to match your workflow needs.

## Agent

The default mode for complex coding tasks. Agent autonomously explores your codebase, edits multiple files, runs commands, and fixes errors to complete your requests.

## Ask

Read-only mode for learning and exploration. Ask searches your codebase and provides answers without making any changes - perfect for understanding code before modifying it.

## Custom

Create your own modes with specific tool combinations and instructions. Mix and match capabilities to fit your workflow.

<Note>
  Custom modes are in beta. Enable them in `Cursor Settings` → `Chat` → `Custom
    Modes`
</Note>

### Examples

<AccordionGroup>
  <Accordion title="Learn">
    **Tools:** All Search\
    **Instructions:** Focus on explaining concepts thoroughly and ask clarifying questions
  </Accordion>

{" "}

<Accordion title="Refactor">
    **Tools:** Edit & Reapply **Instructions:** Improve code structure without
    adding new functionality
  </Accordion>

{" "}

<Accordion title="Plan">
    **Tools:** Codebase, Read file, Terminal **Instructions:** Create detailed
    implementation plans in `plan.md`
  </Accordion>

<Accordion title="Debug">
    **Tools:** All Search, Terminal, Edit & Reapply\
    **Instructions:** Investigate issues thoroughly before proposing fixes
  </Accordion>
</AccordionGroup>

## Switching modes

* Use the mode picker dropdown in Agent
* Press <Kbd>Cmd+.</Kbd> for quick switching
* Set keyboard shortcuts in [settings](#settings)

## Settings

All modes share common configuration options:

Mode-specific settings:

# Overview
Source: https://docs.cursor.com/en/agent/overview

Assistant for autonomous coding tasks, terminal commands, and code editing

Agent is Cursor's assistant that can complete complex coding tasks independently, run terminal commands, and edit code. Access in sidepane with <Kbd>Cmd+I</Kbd>.

<div className="mt-24 flex flex-col gap-12">
  <Columns className="gap-4">
    <div>
      <h2 className="text-lg font-medium mb-2">
        <a href="/en/agent/modes" className="hover:text-primary transition-colors">
          Modes
        </a>
      </h2>

<p className="text-sm">
        Choose between Agent, Ask, or create custom modes. Each mode has
        different capabilities and tools to match your workflow.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/tools" className="hover:text-primary transition-colors">
          Tools
        </a>
      </h3>

<p className="text-sm">
        Agent uses tools to search, edit, and run commands. From semantic codebase
        search to terminal execution, these tools enable autonomous task
        completion.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/apply" className="hover:text-primary transition-colors">
          Apply Changes
        </a>
      </h3>

<p className="text-sm">
        Integrate AI-suggested code blocks into your codebase. Apply handles
        large-scale changes efficiently while maintaining precision.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/review" className="hover:text-primary transition-colors">
          Review Diffs
        </a>
      </h3>

<p className="text-sm">
        Examine changes before accepting them. Review interface shows additions
        and deletions with color-coded lines for control over modifications.
      </p>
    </div>

<p className="text-sm">
        Run multiple conversations simultaneously with <Kbd>Cmd+T</Kbd>. Each tab
        maintains its own context, history, and model selection.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/chats/checkpoints" className="hover:text-primary transition-colors">
          Checkpoints
        </a>
      </h3>

<p className="text-sm">
        Automatic snapshots track Agent's changes. Restore previous states if
        changes don't work as expected or to try different approaches.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/terminal" className="hover:text-primary transition-colors">
          Terminal Integration
        </a>
      </h3>

<p className="text-sm">
        Agent executes terminal commands, monitors output, and handles multi-step
        processes. Configure auto-run for trusted workflows or require
        confirmation for safety.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/chats/history" className="hover:text-primary transition-colors">
          Chat History
        </a>
      </h3>

<p className="text-sm">
        Access past conversations with <Kbd>Opt Cmd '</Kbd>. Review previous
        discussions, track coding sessions, and reference context from earlier
        chats.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/chats/export" className="hover:text-primary transition-colors">
          Export Chats
        </a>
      </h3>

<p className="text-sm">
        Export conversations to markdown format. Share solutions with team
        members, document decisions, or create knowledge bases from coding
        sessions.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/context/rules" className="hover:text-primary transition-colors">
          Rules
        </a>
      </h3>

<p className="text-sm">
        Define custom instructions for Agent behavior. Rules help maintain coding
        standards, enforce patterns, and personalize how Agent assists with your
        project.
      </p>
    </div>

# Planning
Source: https://docs.cursor.com/en/agent/planning

How Agent plans and manages complex tasks with todos and queuing

Agent can plan ahead and manage complex tasks with structured to-do lists and message queuing, making long-horizon tasks easier to understand and track.

## Agent to-dos

Agent can break down longer tasks into manageable steps with dependencies, creating a structured plan that updates as work progresses.

### How it works

* Agent automatically creates to-do lists for complex tasks
* Each item can have dependencies on other tasks
* The list updates in real-time as work progresses
* Completed tasks are marked off automatically

### Visibility

* To-dos appear in the chat interface
* If [Slack integration](/en/slack) is set up, to-dos are also visible there
* You can view the full task breakdown at any time

<Tip>
  For better planning, describe your end goal clearly. Agent will create more
  accurate task breakdowns when it understands the full scope.
</Tip>

<Note>Planning and to-dos are currently not supported for auto mode.</Note>

## Queued messages

Queue follow-up messages while Agent is working on the current task. Your instructions wait in line and execute automatically when ready.

### Using the queue

1. While Agent is working, type your next instruction
2. Press <Kbd>Ctrl+Enter</Kbd> to add it to the queue
3. Messages appear in order below the active task
4. Reorder queued messages by clicking arrow
5. Agent processes them sequentially after finishing

### Override the queue

To queue your message instead of using default messaging, use <Kbd>Ctrl+Enter</Kbd>. To send a message immediately without queuing, use <Kbd>Cmd+Enter</Kbd>. This "force pushes" your message, bypassing the queue to execute right away.

## Default messaging

Messages send as fast as possible by default, typically appearing right after Agent completes a tool call. This creates the most responsive experience.

### How default messaging works

* Your message gets appended to the most recent user message in the chat
* Messages typically attach to tool results and send immediately when ready
* This creates a more natural conversation flow without interrupting Agent's current work
* By default, this happens when you press Enter while Agent is working

# Diffs & Review
Source: https://docs.cursor.com/en/agent/review

Review and manage code changes generated by the AI agent

When Agent generates code changes, they're presented in a review interface that shows additions and deletions with color-coded lines. This allows you to examine and control which changes are applied to your codebase.

The review interface displays code changes in a familiar diff format:

## Diffs

<div className="full-width-table">
  | Type              | Meaning                    | Example                                                                                               |
  | :---------------- | :------------------------- | :---------------------------------------------------------------------------------------------------- |
  | **Added lines**   | New code additions         | <code className="bg-green-100 text-green-800 px-2 py-1 rounded">+ const newVariable = 'hello';</code> |
  | **Deleted lines** | Code removals              | <code className="bg-red-100 text-red-800 px-2 py-1 rounded">- const oldVariable = 'goodbye';</code>   |
  | **Context lines** | Unchanged surrounding code | <code className="bg-gray-100 text-gray-600 px-2 py-1 rounded"> function example() {}</code>           |
</div>

## Review

After generation completes, you'll see a prompt to review all changes before proceeding. This gives you an overview of what will be modified.

### File-by-file

A floating review bar appears at the bottom of your screen, allowing you to:

* **Accept** or **reject** changes for the current file
* Navigate to the **next file** with pending changes
  <Frame>
    <video src="https://mintcdn.com/cursor/BfJOqJ1Wb8EvuXyr/images/chat/review/review-bar.mp4?fit=max&auto=format&n=BfJOqJ1Wb8EvuXyr&q=85&s=5dca0fe7aba3c79e6760cb264821a617" autoPlay loop muted playsInline controls data-path="images/chat/review/review-bar.mp4">
      Your browser does not support the video tag.
    </video>
  </Frame>

### Selective acceptance

For fine-grained control:

* To accept most changes: reject unwanted lines, then click **Accept all**
* To reject most changes: accept wanted lines, then click **Reject all**

## Review changes

At the end of the agent response, click the **Review changes** button to see the full diff of the changes.

# Terminal
Source: https://docs.cursor.com/en/agent/terminal

Run terminal commands automatically as part of agent operations

Agent executes commands in Cursor's native terminal with preserved history. Click skip to send <kbd>Ctrl+C</kbd> and interrupt commands.

## Troubleshooting

<Info>
  Some shell themes (for example, Powerlevel9k/Powerlevel10k) can interfere with
  the inline terminal output. If your command output looks truncated or
  misformatted, disable the theme or switch to a simpler prompt when Agent runs.
</Info>

### Disable heavy prompts for Agent sessions

Use the `CURSOR_AGENT` environment variable in your shell config to detect when
the Agent is running and skip initializing fancy prompts/themes.

```zsh  theme={null}
# ~/.zshrc — disable Powerlevel10k when Cursor Agent runs
if [[ -n "$CURSOR_AGENT" ]]; then
  # Skip theme initialization for better compatibility
else
  [[ -r ~/.p10k.zsh ]] && source ~/.p10k.zsh
fi
```

```bash  theme={null}
# ~/.bashrc — fall back to a simple prompt in Agent sessions
if [[ -n "$CURSOR_AGENT" ]]; then
  PS1='\u@\h \W \$ '
fi
```

# Tools
Source: https://docs.cursor.com/en/agent/tools

Tools available to agents for searching, editing, and running code

A list of all tools available to modes within the [Agent](/en/agent/overview), which you can enable or disable when building your own [custom modes](/en/agent/modes#custom).

<Note>
  There is no limit on the number of tool calls Agent can make during a task. Agent will continue using tools as needed to complete your request.
</Note>

## Search

Tools used to search your codebase and the web to find relevant information.

<AccordionGroup>
  <Accordion title="Read File" icon="file-lines">
    Reads up to 250 lines (750 in max mode) of a file.
  </Accordion>

<Accordion title="List Directory" icon="folder-open">
    Read the structure of a directory without reading file contents.
  </Accordion>

<Accordion title="Codebase" icon="database">
    Perform semantic searches within your [indexed
    codebase](/en/context/codebase-indexing).
  </Accordion>

<Accordion title="Grep" icon="magnifying-glass">
    Search for exact keywords or patterns within files.
  </Accordion>

<Accordion title="Search Files" icon="file-magnifying-glass">
    Find files by name using fuzzy matching.
  </Accordion>

<Accordion title="Web" icon="globe">
    Generate search queries and perform web searches.
  </Accordion>

<Accordion title="Fetch Rules" icon="gavel">
    Retrieve specific [rules](/en/context/rules) based on type and description.
  </Accordion>
</AccordionGroup>

## Edit

Tools used to make specific edits to your files and codebase.

<AccordionGroup>
  <Accordion title="Edit & Reapply" icon="pencil">
    Suggest edits to files and [apply](/en/agent/apply) them automatically.
  </Accordion>

<Accordion title="Delete File" icon="trash">
    Delete files autonomously (can be disabled in settings).
  </Accordion>
</AccordionGroup>

## Run

Chat can interact with your terminal.

<AccordionGroup>
  <Accordion title="Terminal" icon="terminal">
    Execute terminal commands and monitor output.
  </Accordion>
</AccordionGroup>

<Note>By default, Cursor uses the first terminal profile available.</Note>

To set your preferred terminal profile:

1. Open Command Palette (`Cmd/Ctrl+Shift+P`)
2. Search for "Terminal: Select Default Profile"
3. Choose your desired profile

## MCP

Chat can use configured MCP servers to interact with external services, such as databases or 3rd party APIs.

<AccordionGroup>
  <Accordion title="Toggle MCP Servers" icon="server">
    Toggle available MCP servers. Respects auto-run configuration.
  </Accordion>
</AccordionGroup>

Learn more about [Model Context Protocol](/en/context/model-context-protocol) and explore available servers in the [MCP directory](/en/tools).

## Advanced options

<AccordionGroup>
  <Accordion title="Auto-apply Edits" icon="check">
    Automatically apply edits without manual confirmation.
  </Accordion>

<Accordion title="Auto-run" icon="play">
    Automatically execute terminal commands and accept edits. Useful for running test suites and verifying changes.

<Accordion title="Guardrails" icon="shield">
    Configure allow lists to specify which tools can execute automatically. Allow lists provide better security by explicitly defining permitted operations.
  </Accordion>

<Accordion title="Auto-fix Errors" icon="wrench">
    Automatically resolve linter errors and warnings when encountered by Agent.
  </Accordion>
</AccordionGroup>

# Background Agents
Source: https://docs.cursor.com/en/background-agent

Asynchronous remote agents in Cursor

With background agents, spawn asynchronous agents that edit and run code in a remote environment. View their status, send follow-ups, or take over anytime.

## How to Use

You can access background agents in two ways:

1. **Background Agent Sidebar**: Use the background agent tab in the native Cursor sidebar to view all background agents associated with your account, search existing agents, and start new ones.
2. **Background Agent Mode**: Hit <Kbd tooltip="Trigger background agent mode">Cmd E</Kbd> to trigger background agent mode in the UI.

After submitting a prompt, select your agent from the list to view status and enter the machine.

<Note>
  <p className="!mb-0">
    Background agents require data retention on the order of a few days.
  </p>
</Note>

## Setup

Background agents run in an isolated ubuntu-based machine by default. Agents have internet access and can install packages.

#### GitHub connection

Background agents clone your repo from GitHub and work on a separate branch, pushing to your repo for easy handoff.

Grant read-write privileges to your repo (and any dependent repos or submodules). We'll support other providers (GitLab, BitBucket, etc) in the future.

##### IP Allow List Configuration

If your organization uses GitHub's IP allow list feature, you'll need to configure access for background agents. See the [GitHub integration documentation](/en/integrations/github#ip-allow-list-configuration) for complete setup instructions including contact information and IP addresses.

#### Base Environment Setup

For advanced cases, set up the environment yourself. Get an IDE instance connected to the remote machine. Set up your machine, install tools and packages, then take a snapshot. Configure runtime settings:

* Install command runs before an agent starts and installs runtime dependencies. This might mean running `npm install` or `bazel build`.
* Terminals run background processes while the agent works - like starting a web server or compiling protobuf files.

For the most advanced cases, use a Dockerfile for machine setup. The dockerfile lets you set up system-level dependencies: install specific compiler versions, debuggers, or switch the base OS image. Don't `COPY` the entire project - we manage the workspace and check out the correct commit. Still handle dependency installation in the install script.

Enter any required secrets for your dev environment - they're stored encrypted-at-rest (using KMS) in our database and provided in the background agent environment.

The machine setup lives in `.cursor/environment.json`, which can be committed in your repo (recommended) or stored privately. The setup flow guides you through creating `environment.json`.

#### Maintenance Commands

When setting up a new machine, we start from the base environment, then run the `install` command from your `environment.json`. This command is what a developer would run when switching branches - install any new dependencies.

For most people, the `install` command is `npm install` or `bazel build`.

To ensure fast machine startup, we cache disk state after the `install` command runs. Design it to run multiple times. Only disk state persists from the `install` command - processes started here won't be alive when the agent starts.

#### Startup Commands

After running `install`, the machine starts and we run the `start` command followed by starting any `terminals`. This starts processes that should be alive when the agent runs.

The `start` command can often be skipped. Use it if your dev environment relies on docker - put `sudo service docker start` in the `start` command.

`terminals` are for app code. These terminals run in a `tmux` session available to you and the agent. For example, many website repos put `npm run watch` as a terminal.

#### The `environment.json` Spec

The `environment.json` file can look like:

```json  theme={null}
{
  "snapshot": "POPULATED_FROM_SETTINGS",
  "install": "npm install",
  "terminals": [
    {
      "name": "Run Next.js",
      "command": "npm run dev"
    }
  ]
}
```

Formally, the spec is [defined here](https://www.cursor.com/schemas/environment.schema.json).

## Models

Only [Max Mode](/en/context/max-mode)-compatible models are available for background agents.

## Pricing

Learn more about [Background Agent pricing](/en/account/pricing#background-agent).

## Security

Background Agents are available in Privacy Mode. We never train on your code and only retain code for running the agent. [Learn more about Privacy mode](https://www.cursor.com/privacy-overview).

What you should know:

1. Grant read-write privileges to our GitHub app for repos you want to edit. We use this to clone the repo and make changes.
2. Your code runs inside our AWS infrastructure in isolated VMs and is stored on VM disks while the agent is accessible.
3. The agent has internet access.
4. The agent auto-runs all terminal commands, letting it iterate on tests. This differs from the foreground agent, which requires user approval for every command. Auto-running introduces data exfiltration risk: attackers could execute prompt injection attacks, tricking the agent to upload code to malicious websites. See [OpenAI's explanation about risks of prompt injection for background agents](https://platform.openai.com/docs/codex/agent-network#risks-of-agent-internet-access).
5. If privacy mode is disabled, we collect prompts and dev environments to improve the product.
6. If you disable privacy mode when starting a background agent, then enable it during the agent's run, the agent continues with privacy mode disabled until it completes.

## Dashboard settings

Workspace admins can configure additional settings from the Background Agents tab on the dashboard.

### Defaults Settings

* **Default model** – the model used when a run does not specify one. Pick any model that supports Max Mode.
* **Default repository** – when empty, agents ask the user to choose a repo. Supplying a repo here lets users skip that step.
* **Base branch** – the branch agents fork from when creating pull requests. Leave blank to use the repository’s default branch.

### Security Settings

All security options require admin privileges.

* **User restrictions** – choose *None* (all members can start background agents) or *Allow list*. When set to *Allow list* you specify exactly which teammates can create agents.
* **Team follow-ups** – when on, anyone in the workspace can add follow-up messages to an agent someone else started. Turn it off to restrict follow-ups to the agent owner and admins.
* **Display agent summary** – controls whether Cursor shows the agent’s file-diff images and code snippets. Disable this if you prefer not to expose file paths or code in the sidebar.
* **Display agent summary in external channels** – extends the previous toggle to Slack or any external channel you’ve connected.

Changes save instantly and affect new agents immediately.

# Add Follow-up
Source: https://docs.cursor.com/en/background-agent/api/add-followup

en/background-agent/api/openapi.yaml post /v0/agents/{id}/followup
Send an additional instruction to a running background agent.

# Agent Conversation
Source: https://docs.cursor.com/en/background-agent/api/agent-conversation

en/background-agent/api/openapi.yaml get /v0/agents/{id}/conversation
Retrieve the conversation history of a background agent.

If the background agent has been deleted, you cannot access the conversation.

# Agent Status
Source: https://docs.cursor.com/en/background-agent/api/agent-status

en/background-agent/api/openapi.yaml get /v0/agents/{id}
Get the current status and results of a specific background agent.

# API Key Info
Source: https://docs.cursor.com/en/background-agent/api/api-key-info

en/background-agent/api/openapi.yaml get /v0/me
Retrieve metadata about the API key used for authentication.

# Delete an Agent
Source: https://docs.cursor.com/en/background-agent/api/delete-agent

en/background-agent/api/openapi.yaml delete /v0/agents/{id}
Permanently delete a background agent and its associated resources.

# Launch an Agent
Source: https://docs.cursor.com/en/background-agent/api/launch-an-agent

en/background-agent/api/openapi.yaml post /v0/agents
Start a new background agent to work on your repository.

# List Agents
Source: https://docs.cursor.com/en/background-agent/api/list-agents

en/background-agent/api/openapi.yaml get /v0/agents
Retrieve a paginated list of all background agents for the authenticated user.

# List Models
Source: https://docs.cursor.com/en/background-agent/api/list-models

en/background-agent/api/openapi.yaml get /v0/models
Retrieve a list of recommended models for background agents.

If you want to provide the background agent's model during creation, you can use this endpoint to see a list of recommended models.

In that case, we also recommend having an "Auto" option, in which you would not provide a model name to the creation endpoint,
and we will pick the most appropriate model.

# List GitHub Repositories
Source: https://docs.cursor.com/en/background-agent/api/list-repositories

en/background-agent/api/openapi.yaml get /v0/repositories
Retrieve a list of GitHub repositories accessible to the authenticated user.

<Warning>
  **This endpoint has very strict rate limits.**

Limit requests to **1 / user / minute**, and **30 / user / hour.**

This request can take tens of seconds to respond for users with access to many repositories.

Make sure to handle this information not being available gracefully.
</Warning>

# Overview
Source: https://docs.cursor.com/en/background-agent/api/overview

Programmatically create and manage background agents that work on your repositories

# Background Agents API

The Background Agents API allows you to programmatically create and manage AI-powered coding agents that work autonomously on your repositories.
You can use the API to automatically respond to user feedback, fix bugs, update docs, and much more!

<Info>
  Background Agents API is currently in beta, we'd love your feedback on it!
</Info>

## Key features

* **Autonomous code generation** - Create agents that can understand your prompt and make changes to your codebase
* **Repository integration** - Work directly with GitHub repositories
* Follow-up prompts - Add additional instructions to running agents
* **Usage-based pricing** - Pay only for the tokens you use
* **Scalable** - Support for up to 256 active agents per API key

## Quick start

### 1. Get your API key

**Navigate** to [Cursor Dashboard → Integrations](https://cursor.com/dashboard?tab=integrations) to create your API key.

### 2. Start using the API

All API endpoints are relative to:

```
https://api.cursor.com
```

See the [API reference](/en/background-agent/api/launch-an-agent) for a detailed list of endpoints.

## Authentication

All API requests require authentication using a Bearer token:

```
Authorization: Bearer YOUR_API_KEY
```

API keys are created in the [Cursor Dashboard](https://cursor.com/dashboard?tab=integrations). Keys are scoped to your account and grant permission to create and manage agents (subject to your plan limits and repository access).

## Pricing

The API is currently in beta with the same pricing as Background Agents. Pricing may change as we scale the service. See [Background Agent pricing](/en/account/pricing#background-agent).

## Next steps

* Read the main [Background Agents overview](/en/background-agent) to understand environments, permissions, and workflows.
* Try Background Agents from [web & mobile](/en/background-agent/web-and-mobile).
* Join the discussion in [Discord #background-agent](https://discord.gg/jfgpZtYpmb) or email [background-agent-feedback@cursor.com](mailto:background-agent-feedback@cursor.com).

# Webhooks
Source: https://docs.cursor.com/en/background-agent/api/webhooks

Receive real-time notifications about background agent status changes

# Webhooks

When you create an agent with a webhook URL, Cursor will send HTTP POST requests to notify you about status changes. Currently, only `statusChange` events are supported, specifically when an agent encounters an `ERROR` or `FINISHED` state.

## Webhook verification

To ensure the webhook requests are authentically from Cursor, verify the signature included with each request:

### Headers

Each webhook request includes the following headers:

* **`X-Webhook-Signature`** – Contains the HMAC-SHA256 signature in the format `sha256=<hex_digest>`
* **`X-Webhook-ID`** – A unique identifier for this delivery (useful for logging)
* **`X-Webhook-Event`** – The event type (currently only `statusChange`)
* **`User-Agent`** – Always set to `Cursor-Agent-Webhook/1.0`

### Signature verification

To verify the webhook signature, compute the expected signature and compare it with the received signature:

```javascript  theme={null}
const crypto = require('crypto');

function verifyWebhook(secret, rawBody, signature) {
  const expectedSignature = 'sha256=' + 
    crypto.createHmac('sha256', secret)
          .update(rawBody)
          .digest('hex');
  
  return signature === expectedSignature;
}
```

```python  theme={null}
import hmac
import hashlib

def verify_webhook(secret, raw_body, signature):
    expected_signature = 'sha256=' + hmac.new(
        secret.encode(),
        raw_body,
        hashlib.sha256
    ).hexdigest()
    
    return signature == expected_signature
```

Always use the raw request body (before any parsing) when computing the signature.

## Payload format

The webhook payload is sent as JSON with the following structure:

```json  theme={null}
{
  "event": "statusChange",
  "timestamp": "2024-01-15T10:30:00Z",
  "id": "bc_abc123",
  "status": "FINISHED",
  "source": {
    "repository": "https://github.com/your-org/your-repo",
    "ref": "main"
  },
  "target": {
    "url": "https://cursor.com/agents?id=bc_abc123",
    "branchName": "cursor/add-readme-1234",
    "prUrl": "https://github.com/your-org/your-repo/pull/1234"
  },
  "summary": "Added README.md with installation instructions"
}
```

Note that some fields are optional and will only be included when available.

## Best practices

* **Verify signatures** – Always verify the webhook signature to ensure the request is from Cursor
* **Handle retries** – Webhooks may be retried if your endpoint returns an error status code
* **Return quickly** – Return a 2xx status code as soon as possible
* **Use HTTPS** – Always use HTTPS URLs for webhook endpoints in production
* **Store raw payloads** – Store the raw webhook payload for debugging and future verification

# Web & Mobile
Source: https://docs.cursor.com/en/background-agent/web-and-mobile

Run coding agents from any device with seamless handoff to desktop

## Overview

Cursor's Agent on web brings a powerful coding assistant to every device. Whether you're on your phone during a walk, or working in your web browser, you can now kick off powerful coding agents that work in the background.
When they're done, pick up their work inside Cursor, review and merge changes, or share links with your team to collaborate.

Get started at [cursor.com/agents](https://cursor.com/agents).

## Getting started

### Quick setup

1. **Visit the web app**: Go to [cursor.com/agents](https://cursor.com/agents) on any device
2. **Sign in**: Log in with your Cursor account
3. **Connect GitHub**: Link your GitHub account to access repositories
4. **Start your first agent**: Type in a task and watch the agent get to work

### Mobile installation

For the best mobile experience, install Cursor as a Progressive Web App (PWA):

* **iOS**: Open [cursor.com/agents](https://cursor.com/agents) in Safari, tap the share button, then "Add to Home Screen"
* **Android**: Open the URL in Chrome, tap the menu, then "Add to Home Screen" or "Install App"

<Tip>
  Installing as a PWA provides a native-feeling experience with: - Full-screen
  interface - Faster startup times - App icon on your home screen
</Tip>

## Working across devices

The Web and Mobile Agent is designed to work with your desktop workflow; click "Open in Cursor" to continue the agent's work in your IDE.

### Team collaboration

* **Shared access**: Share links with team members to collaborate on agent runs.
* **Review process**: Collaborators can review diffs and provide feedback.
* **Pull request management**: Create, review, and merge pull requests directly from the web interface.

### Slack integration

Trigger agents directly from Slack by mentioning `@Cursor`, and when starting agents from web or mobile, opt to receive Slack notifications upon completion.

<Card title="Use Cursor in Slack" icon="slack" href="/en/slack">
  Learn more about setting up and using the Slack integration, including
  triggering agents and receiving notifications.
</Card>

## Pricing

Web & mobile agents use the same pricing model as Background Agents.

Learn more about [Background Agent pricing](/en/account/pricing#background-agent).

## Troubleshooting

<AccordionGroup>
  <Accordion title="Agent runs are not starting">
    * Ensure you're logged in and have connected your GitHub account. - Check
      that you have the necessary repository permissions - You will also need to
      be on a Pro Trial or paid plan with usage based pricing enabled. To enable
      usage based pricing, go to your
      [Dashboard](https://www.cursor.com/dashboard?tab=settings) settings tab.
  </Accordion>

<Accordion title="Can't see agent runs on mobile">
    Try refreshing the page or clearing your browser cache. Ensure you're using
    the same account across devices.
  </Accordion>

<Accordion title="Slack integration not working">
    Verify that your workspace admin has installed the Cursor Slack app and that
    you have the proper permissions.
  </Accordion>
</AccordionGroup>

# Bugbot
Source: https://docs.cursor.com/en/bugbot

AI code review for pull requests

Bugbot reviews pull requests and identifies bugs, security issues, and code quality problems.

<Tip>
  Bugbot includes a free tier: every user gets a limited number of free PR reviews each month. When you reach the limit, reviews pause until your next billing cycle. You can upgrade anytime to a 14‑day free Pro trial for unlimited reviews (subject to standard abuse guardrails).
</Tip>

## How it works

Bugbot analyzes PR diffs and leaves comments with explanations and fix suggestions. It runs automatically on each PR update or manually when triggered.

* Runs **automatic reviews** on every PR update
* **Manual trigger** by commenting `cursor review` or `bugbot run` on any PR
* **Fix in Cursor** links open issues directly in Cursor
* **Fix in Web** links open issues directly in [cursor.com/agents](https://cursor.com/agents)

## Setup

Requires Cursor admin access and GitHub org admin access.

1. Go to [cursor.com/dashboard](https://cursor.com/dashboard?tab=bugbot)
2. Navigate to the Bugbot tab
3. Click `Connect GitHub` (or `Manage Connections` if already connected)
4. Follow the GitHub installation flow
5. Return to the dashboard to enable Bugbot on specific repositories

## Configuration

<Tabs defaultValue="Team">
  <Tab title="Individual">
    ### Repository settings

Enable or disable Bugbot per repository from your installations list. Bugbot runs only on PRs you author.

### Personal settings

* Run **only when mentioned** by commenting `cursor review` or `bugbot run`
    * Run **only once** per PR, skipping subsequent commits
  </Tab>

<Tab title="Team">
    ### Repository settings

Team admins can enable Bugbot per repository, configure allow/deny lists for reviewers, and set:

* Run **only once** per PR per installation, skipping subsequent commits
    * **Disable inline reviews** to prevent Bugbot from leaving comments directly on code lines

Bugbot runs for all contributors to enabled repositories, regardless of team membership.

### Personal settings

Team members can override settings for their own PRs:

* Run **only when mentioned** by commenting `cursor review` or `bugbot run`
    * Run **only once** per PR, skipping subsequent commits
    * **Enable reviews on draft PRs** to include draft pull requests in automatic reviews
  </Tab>
</Tabs>

### Analytics

## Rules

Create `.cursor/BUGBOT.md` files to provide project-specific context for reviews. Bugbot always includes the root `.cursor/BUGBOT.md` file and any additional files found while traversing upward from changed files.

```
project/
  .cursor/BUGBOT.md          # Always included (project-wide rules)
  backend/
    .cursor/BUGBOT.md        # Included when reviewing backend files
    api/
      .cursor/BUGBOT.md      # Included when reviewing API files
  frontend/
    .cursor/BUGBOT.md        # Included when reviewing frontend files
```

<AccordionGroup>
  <Accordion title="Example .cursor/BUGBOT.md">
    ```markdown  theme={null}
    # Project review guidelines

## Security focus areas

- Validate user input in API endpoints
    - Check for SQL injection vulnerabilities in database queries
    - Ensure proper authentication on protected routes

## Architecture patterns

- Use dependency injection for services
    - Follow the repository pattern for data access
    - Implement proper error handling with custom error classes

## Common issues

- Memory leaks in React components (check useEffect cleanup)
    - Missing error boundaries in UI components
    - Inconsistent naming conventions (use camelCase for functions)

```
  </Accordion>
</AccordionGroup>

## Pricing

Bugbot offers two tiers: **Free** and **Pro**.

### Free tier

Every user gets a limited number of free PR reviews each month. For teams, each team member gets their own free reviews. When you reach the limit, reviews pause until your next billing cycle. You can upgrade anytime to the 14‑day free Pro trial for unlimited reviews.

### Pro tier

<Tabs defaultValue="Teams">
  <Tab title="Individuals">
    ### Flat rate

\$40 per month for unlimited Bugbot reviews on up to 200 PRs per month across all repositories.

### Getting started

Subscribe through your account settings.
  </Tab>

<Tab title="Teams">
    ### Per-user billing

Teams pay \$40 per user per month for unlimited reviews.

We count a user as someone who authored PRs reviewed by Bugbot in a month.

All licenses are relinquished at the start of each billing cycle, and will be assigned out on a first-come, first-served basis. If a user doesn't author any PRs reviewed by Bugbot in a month, the seat can be used by another user.

### Seat limits

Team admins can set maximum Bugbot seats per month to control costs.

### Getting started

Subscribe through your team dashboard to enable billing.

### Abuse guardrails

In order to prevent abuse, we have a pooled cap of 200 pull requests per month for every Bugbot license. If you need more than 200 pull requests per month, please contact us at [hi@cursor.com](mailto:hi@cursor.com) and we'll be happy to help you out.

For example, if your team has 100 users, your organization will initially be able to review 20,000 pull requests per month. If you reach that limit naturally, please reach out to us and we'll be happy to increase the limit.
  </Tab>
</Tabs>

## Troubleshooting

If Bugbot isn't working:

1. **Enable verbose mode** by commenting `cursor review verbose=true` or `bugbot run verbose=true` for detailed logs and request ID
2. **Check permissions** to verify Bugbot has repository access
3. **Verify installation** to confirm the GitHub app is installed and enabled

Include the request ID from verbose mode when reporting issues.

## FAQ

<AccordionGroup>
  <Accordion title="Is Bugbot privacy-mode compliant?">
    Yes, Bugbot follows the same privacy compliance as Cursor and processes data identically to other Cursor requests.
  </Accordion>

<Accordion title="What happens when I hit the free tier limit?">
    When you reach your monthly free tier limit, Bugbot reviews pause until your next billing cycle. You can upgrade to the 14‑day free Pro trial for unlimited reviews (subject to standard abuse guardrails).
  </Accordion>
</AccordionGroup>

```
```

# Code Review
Source: https://docs.cursor.com/en/cli/cookbook/code-review

Build a GitHub Actions workflow that uses Cursor CLI to automatically review pull requests and provide feedback

This tutorial shows you how to set up code review using Cursor CLI in GitHub Actions. The workflow will analyze pull requests, identify issues, and post feedback as comments.

<Tip>
  For most users, we recommend using [Bugbot](/en/bugbot) instead. Bugbot provides managed automated code review with no setup required. This CLI approach is useful to explore capabilities and for advanced customization.
</Tip>

<div className="space-y-4">
  <Expandable title="full workflow file">
    ```yaml cursor-code-review.yml theme={null}
    name: Code Review

on:
      pull_request:
        types: [opened, synchronize, reopened, ready_for_review]

permissions:
      pull-requests: write
      contents: read
      issues: write

jobs:
      code-review:
        runs-on: ubuntu-latest
        # Skip automated code review for draft PRs
        if: github.event.pull_request.draft == false
        steps:
          - name: Checkout repository
            uses: actions/checkout@v4
            with:
              fetch-depth: 0
              ref: ${{ github.event.pull_request.head.sha }}

- name: Install Cursor CLI
            run: |
              curl https://cursor.com/install -fsS | bash
              echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
            run: |
              git config user.name "Cursor Agent"
              git config user.email "cursoragent@cursor.com"

- name: Perform automated code review
            env:
              CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
              MODEL: gpt-5
              GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
              BLOCKING_REVIEW: ${{ vars.BLOCKING_REVIEW || 'false' }}
            run: |
              cursor-agent --force --model "$MODEL" --output-format=text --print 'You are operating in a GitHub Actions runner performing automated code review. The gh CLI is available and authenticated via GH_TOKEN. You may comment on pull requests.

Context:
              - Repo: ${{ github.repository }}
              - PR Number: ${{ github.event.pull_request.number }}
              - PR Head SHA: ${{ github.event.pull_request.head.sha }}
              - PR Base SHA: ${{ github.event.pull_request.base.sha }}
              - Blocking Review: ${{ env.BLOCKING_REVIEW }}

Objectives:
              1) Re-check existing review comments and reply resolved when addressed.
              2) Review the current PR diff and flag only clear, high-severity issues.
              3) Leave very short inline comments (1-2 sentences) on changed lines only and a brief summary at the end.

Procedure:
              - Get existing comments: gh pr view --json comments
              - Get diff: gh pr diff
              - Get changed files with patches to compute inline positions: gh api repos/${{ github.repository }}/pulls/${{ github.event.pull_request.number }}/files --paginate --jq '.[] | {filename,patch}'
              - Compute exact inline anchors for each issue (file path + diff position). Comments MUST be placed inline on the changed line in the diff, not as top-level comments.
              - Detect prior top-level "no issues" style comments authored by this bot (match bodies like: "✅ no issues", "No issues found", "LGTM").
              - If CURRENT run finds issues and any prior "no issues" comments exist:
                - Prefer to remove them to avoid confusion:
                  - Try deleting top-level issue comments via: gh api -X DELETE repos/${{ github.repository }}/issues/comments/<comment_id>
                  - If deletion isn't possible, minimize them via GraphQL (minimizeComment) or edit to prefix "[Superseded by new findings]".
                - If neither delete nor minimize is possible, reply to that comment: "⚠️ Superseded: issues were found in newer commits".
              - If a previously reported issue appears fixed by nearby changes, reply: ✅ This issue appears to be resolved by the recent changes
              - Analyze ONLY for:
                - Null/undefined dereferences
                - Resource leaks (unclosed files or connections)
                - Injection (SQL/XSS)
                - Concurrency/race conditions
                - Missing error handling for critical operations
                - Obvious logic errors with incorrect behavior
                - Clear performance anti-patterns with measurable impact
                - Definitive security vulnerabilities
              - Avoid duplicates: skip if similar feedback already exists on or near the same lines.

Commenting rules:
              - Max 10 inline comments total; prioritize the most critical issues
              - One issue per comment; place on the exact changed line
              - All issue comments MUST be inline (anchored to a file and line/position in the PR diff)
              - Natural tone, specific and actionable; do not mention automated or high-confidence
              - Use emojis: 🚨 Critical 🔒 Security ⚡ Performance ⚠️ Logic ✅ Resolved ✨ Improvement

Submission:
              - If there are NO issues to report and an existing top-level comment indicating "no issues" already exists (e.g., "✅ no issues", "No issues found", "LGTM"), do NOT submit another comment. Skip submission to avoid redundancy.
              - If there are NO issues to report and NO prior "no issues" comment exists, submit one brief summary comment noting no issues.
              - If there ARE issues to report and a prior "no issues" comment exists, ensure that prior comment is deleted/minimized/marked as superseded before submitting the new review.
              - If there ARE issues to report, submit ONE review containing ONLY inline comments plus an optional concise summary body. Use the GitHub Reviews API to ensure comments are inline:
                - Build a JSON array of comments like: [{ "path": "<file>", "position": <diff_position>, "body": "..." }]
                - Submit via: gh api repos/${{ github.repository }}/pulls/${{ github.event.pull_request.number }}/reviews -f event=COMMENT -f body="$SUMMARY" -f comments='[$COMMENTS_JSON]'
              - Do NOT use: gh pr review --approve or --request-changes

Blocking behavior:
              - If BLOCKING_REVIEW is true and any 🚨 or 🔒 issues were posted: echo "CRITICAL_ISSUES_FOUND=true" >> $GITHUB_ENV
              - Otherwise: echo "CRITICAL_ISSUES_FOUND=false" >> $GITHUB_ENV
              - Always set CRITICAL_ISSUES_FOUND at the end
              '

- name: Check blocking review results
            if: env.BLOCKING_REVIEW == 'true'
            run: |
              echo "Checking for critical issues..."
              echo "CRITICAL_ISSUES_FOUND: ${CRITICAL_ISSUES_FOUND:-unset}"

if [ "${CRITICAL_ISSUES_FOUND:-false}" = "true" ]; then
                echo "❌ Critical issues found and blocking review is enabled. Failing the workflow."
                exit 1
              else
                echo "✅ No blocking issues found."
              fi
    ```
  </Expandable>

## Configure authentication

[Set up your API key and repository secrets](/en/cli/github-actions#authentication) to authenticate Cursor CLI in GitHub Actions.

## Set up agent permissions

Create a configuration file to control what actions the agent can perform. This prevents unintended operations like pushing code or creating pull requests.

Create `.cursor/cli.json` in your repository root:

```json  theme={null}
{
  "permissions": {
    "deny": [
      "Shell(git push)",
      "Shell(gh pr create)",
      "Write(**)"
    ]
  }
}
```

This configuration allows the agent to read files and use the GitHub CLI for comments, but prevents it from making changes to your repository. See the [permissions reference](/en/cli/reference/permissions) for more configuration options.

## Build the GitHub Actions workflow

Now let's build the workflow step by step.

### Set up the workflow trigger

Create `.github/workflows/cursor-code-review.yml` and configure it to run on pull requests:

```yaml  theme={null}
name: Cursor Code Review

on:
  pull_request:
    types: [opened, synchronize, reopened, ready_for_review]

jobs:
  code-review:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write
    
    steps:
```

### Checkout the repository

Add the checkout step to access the pull request code:

```yaml  theme={null}
- name: Checkout repository
  uses: actions/checkout@v4
  with:
    fetch-depth: 0
    ref: ${{ github.event.pull_request.head.sha }}
```

### Install Cursor CLI

Add the CLI installation step:

```yaml  theme={null}
- name: Install Cursor CLI
  run: |
    curl https://cursor.com/install -fsS | bash
    echo "$HOME/.cursor/bin" >> $GITHUB_PATH
```

### Configure the review agent

Before implementing the full review step, let's understand the anatomy of our review prompt. This section outlines how we want the agent to behave:

**Objective**:
We want the agent to review the current PR diff and flag only clear, high-severity issues, then leave very short inline comments (1-2 sentences) on changed lines only with a brief summary at the end. This keeps the signal-to-noise ratio balanced.

**Format**:
We want comments that are short and to the point. We use emojis to make scanning comments easier, and we want a high-level summary of the full review at the end.

**Submission**:
When the review is done, we want the agent to include a short comment based on what was found during the review. The agent should submit one review containing inline comments plus a concise summary.

**Edge cases**:
We need to handle:

* Existing comments being resolved: The agent should mark them as done when addressed
* Avoid duplicates: The agent should skip commenting if similar feedback already exists on or near the same lines

**Final prompt**:
The complete prompt combines all these behavioral requirements to create focused, actionable feedback

Now let's implement the review agent step:

```yaml  theme={null}
- name: Perform code review
  env:
    CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
    GH_TOKEN: ${{ github.token }}
  run: |
    cursor-agent --force --model "$MODEL" --output-format=text --print "You are operating in a GitHub Actions runner performing automated code review. The gh CLI is available and authenticated via GH_TOKEN. You may comment on pull requests.
    
    Context:
    - Repo: ${{ github.repository }}
    - PR Number: ${{ github.event.pull_request.number }}
    - PR Head SHA: ${{ github.event.pull_request.head.sha }}
    - PR Base SHA: ${{ github.event.pull_request.base.sha }}
    
    Objectives:
    1) Re-check existing review comments and reply resolved when addressed
    2) Review the current PR diff and flag only clear, high-severity issues
    3) Leave very short inline comments (1-2 sentences) on changed lines only and a brief summary at the end
    
    Procedure:
    - Get existing comments: gh pr view --json comments
    - Get diff: gh pr diff
    - If a previously reported issue appears fixed by nearby changes, reply: ✅ This issue appears to be resolved by the recent changes
    - Avoid duplicates: skip if similar feedback already exists on or near the same lines
    
    Commenting rules:
    - Max 10 inline comments total; prioritize the most critical issues
    - One issue per comment; place on the exact changed line
    - Natural tone, specific and actionable; do not mention automated or high-confidence
    - Use emojis: 🚨 Critical 🔒 Security ⚡ Performance ⚠️ Logic ✅ Resolved ✨ Improvement
    
    Submission:
    - Submit one review containing inline comments plus a concise summary
    - Use only: gh pr review --comment
    - Do not use: gh pr review --approve or --request-changes"
```

```text  theme={null}
.
├── .cursor/
│   └── cli.json
├── .github/
│   └── workflows/
│       └── cursor-code-review.yml
```

## Test your reviewer

Create a test pull request to verify the workflow works and the agent posts review comments with emoji feedback.

## Next steps

You now have a working automated code review system. Consider these enhancements:

* Set up additional workflows for [fixing CI failures](/en/cli/cookbook/fix-ci)
* Configure different review levels for different branches
* Integrate with your team's existing code review process
* Customize the agent's behavior for different file types or directories

<Expandable title="Advanced: Blocking reviews">
  You can configure the workflow to fail if critical issues are found, preventing the pull request from being merged until addressed.

**Add blocking behavior to the prompt**

First, update your review agent step to include the `BLOCKING_REVIEW` environment variable and add this blocking behavior to the prompt:

```
  Blocking behavior:
  - If BLOCKING_REVIEW is true and any 🚨 or 🔒 issues were posted: echo "CRITICAL_ISSUES_FOUND=true" >> $GITHUB_ENV
  - Otherwise: echo "CRITICAL_ISSUES_FOUND=false" >> $GITHUB_ENV
  - Always set CRITICAL_ISSUES_FOUND at the end
  ```

**Add the blocking check step**

Then add this new step after your code review step:

```yaml  theme={null}
        - name: Check blocking review results
          if: env.BLOCKING_REVIEW == 'true'
          run: |
            echo "Checking for critical issues..."
            echo "CRITICAL_ISSUES_FOUND: ${CRITICAL_ISSUES_FOUND:-unset}"

if [ "${CRITICAL_ISSUES_FOUND:-false}" = "true" ]; then
              echo "❌ Critical issues found and blocking review is enabled. Failing the workflow."
              exit 1
            else
              echo "✅ No blocking issues found."
            fi
  ```
</Expandable>

# Fix CI Failures
Source: https://docs.cursor.com/en/cli/cookbook/fix-ci

Fix CI issues for a repository by using Cursor CLI in GitHub Actions

Fix CI failures using Cursor CLI in GitHub Actions. This workflow analyzes failures, makes targeted fixes, and creates a fix branch with a quick-create PR link.

This workflow monitors a specific workflow by name. Update the `workflows` list to match your actual CI workflow name.

<CodeGroup>
  ```yaml auto-fix-ci.yml theme={null}
  name: Fix CI Failures

on:
    workflow_run:
      workflows: [Test]
      types: [completed]

permissions:
    contents: write
    pull-requests: write
    actions: read

jobs:
    attempt-fix:
      if: >-
        ${{ github.event.workflow_run.conclusion == 'failure' && github.event.workflow_run.name != 'Fix CI Failures' }}
      runs-on: ubuntu-latest
      steps:
        - name: Checkout repository
          uses: actions/checkout@v4
          with:
            fetch-depth: 0

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

- name: Fix CI failure
          env:
            CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
            MODEL: gpt-5
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: ci-fix
          run: |
            cursor-agent -p "You are operating in a GitHub Actions runner.

The GitHub CLI is available as `gh` and authenticated via `GH_TOKEN`. Git is available. You have write access to repository contents and can comment on pull requests, but you must not create or edit PRs directly.

# Context:
            - Repo: ${{ github.repository }}
            - Owner: ${{ github.repository_owner }}
            - Workflow Run ID: ${{ github.event.workflow_run.id }}
            - Workflow Run URL: ${{ github.event.workflow_run.html_url }}
            - Fix Branch Prefix: ${{ env.BRANCH_PREFIX }}

# Goal:
            - Implement an end-to-end CI fix flow driven by the failing PR, creating a separate persistent fix branch and proposing a quick-create PR back into the original PR's branch.

# Requirements:
            1) Identify the PR associated with the failed workflow run and determine its base and head branches. Let HEAD_REF be the PR's head branch (the contributor/origin branch).
            2) Maintain a persistent fix branch for this PR head using the Fix Branch Prefix from Context. Create it if missing, update it otherwise, and push changes to origin.
            3) Attempt to resolve the CI failure by making minimal, targeted edits consistent with the repo's style. Keep changes scoped and safe.
            4) You do NOT have permission to create PRs. Instead, post or update a single natural-language PR comment (1–2 sentences) that briefly explains the CI fix and includes an inline compare link to quick-create a PR.

# Inputs and conventions:
            - Use `gh api`, `gh run view`, `gh pr view`, `gh pr diff`, `gh pr list`, `gh run download`, and git commands as needed to discover the failing PR and branches.
            - Avoid duplicate comments; if a previous bot comment exists, update it instead of posting a new one.
            - If no actionable fix is possible, make no changes and post no comment.

# Deliverables when updates occur:
            - Pushed commits to the persistent fix branch for this PR head.
            - A single natural-language PR comment on the original PR that includes the inline compare link above.
            " --force --model "$MODEL" --output-format=text

```
</CodeGroup>

# Secret Audit
Source: https://docs.cursor.com/en/cli/cookbook/secret-audit

Audit secrets for a repository by using Cursor CLI in GitHub Actions

Audit your repository for security vulnerabilities and secrets exposure using Cursor CLI. This workflow scans for potential secrets, detects risky workflow patterns, and proposes security fixes.

<CodeGroup>
  ```yaml auto-secret-audit.yml theme={null}
  name: Secrets Audit

on:
    schedule:
      - cron: "0 4 * * *"
    workflow_dispatch:

permissions:
    contents: write
    pull-requests: write
    actions: read

jobs:
    secrets-audit:
      runs-on: ubuntu-latest
      steps:
        - name: Checkout repository
          uses: actions/checkout@v4
          with:
            fetch-depth: 0

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

- name: Scan and propose hardening
          env:
            CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
            MODEL: gpt-5
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: audit
          run: |
            cursor-agent -p "You are operating in a GitHub Actions runner.

# Context:
            - Repo: ${{ github.repository }}
             - Hardening Branch Prefix: ${{ env.BRANCH_PREFIX }}

# Goal:
            - Perform a repository secrets exposure and workflow hardening audit on a schedule, and propose minimal safe fixes.

# Requirements:
            1) Scan for potential secrets in tracked files and recent history; support allowlist patterns if present (e.g., .gitleaks.toml).
            2) Detect risky workflow patterns: unpinned actions, overbroad permissions, unsafe pull_request_target usage, secrets in forked PR contexts, deprecated insecure commands, missing permissions blocks.
            3) Maintain a persistent branch for this run using the Hardening Branch Prefix from Context. Create it if missing, update it otherwise, and push changes to origin.
            4) Propose minimal edits: redact literals where safe, add ignore rules, pin actions to SHA, reduce permissions, add guardrails to workflows, and add a SECURITY_LOG.md summarizing changes and remediation guidance.
            5) Push to origin.
            6) If there is at least one open PR in the repo, post or update a single natural-language comment (1–2 sentences) on the most recently updated open PR that briefly explains the hardening changes and includes an inline compare link to quick-create a PR.
            7) Avoid duplicate comments; update an existing bot comment if present. If no changes or no open PRs, post nothing.

# Inputs and conventions:
            - Use `gh` to list PRs and to post comments. Avoid duplicate comments.

# Deliverables when updates occur:
             - Pushed commits to the persistent hardening branch for this run.
            - A single natural-language PR comment with the compare link above (only if an open PR exists).
            " --force --model "$MODEL" --output-format=text

```
</CodeGroup>

# Translate Keys
Source: https://docs.cursor.com/en/cli/cookbook/translate-keys

Translate keys for a repository by using Cursor CLI in GitHub Actions

Manage translation keys for internationalization using Cursor CLI. This workflow detects new or changed i18n keys in pull requests and fills missing translations without overwriting existing ones.

<CodeGroup>
  ```yaml auto-translate-keys.yml theme={null}
  name: Translate Keys

on:
    pull_request:
      types: [opened, synchronize, reopened, ready_for_review]

permissions:
    contents: write
    pull-requests: write

jobs:
    i18n:
      if: ${{ !startsWith(github.head_ref, 'translate/') }}
      runs-on: ubuntu-latest
      steps:
        - name: Checkout repository
          uses: actions/checkout@v4
          with:
            fetch-depth: 0

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

- name: Propose i18n updates
          env:
            CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
            MODEL: gpt-5
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: translate
          run: |
            cursor-agent -p "You are operating in a GitHub Actions runner.

# Context:
            - Repo: ${{ github.repository }}
            - PR Number: ${{ github.event.pull_request.number }}
            - Head Ref: ${{ github.head_ref }}
            - Translate Branch Prefix: ${{ env.BRANCH_PREFIX }}

# Goal:
            - Detect i18n keys added or changed in the PR and fill only missing locales in message files. Never overwrite existing translations.

# Requirements:
            1) Determine changed keys by inspecting the PR diff (source files and messages files).
            2) Compute missing keys per locale using the source/canonical locale as truth.
            3) Add entries only for missing keys. Preserve all existing values untouched.
            4) Validate JSON formatting and schemas.
            5) Maintain a persistent translate branch for this PR head using the Translate Branch Prefix from Context. Create it if missing, update it otherwise, and push changes to origin.
            6) Post or update a single PR comment on the original PR written in natural language (1–2 sentences) that briefly explains what was updated and why, and includes an inline compare link to quick-create a PR.
            7) Avoid duplicate comments; update a previous bot comment if present.
            8) If no changes are necessary, make no commits and post no comment.

# Inputs and conventions:
            - Use `gh pr diff` and git history to detect changes.

# Deliverables when updates occur:
            - Pushed commits to the persistent translate branch for this PR head.
            - A single natural-language PR comment on the original PR with the compare link above.
            " --force --model "$MODEL" --output-format=text

```
</CodeGroup>

# Update Docs
Source: https://docs.cursor.com/en/cli/cookbook/update-docs

Update docs for a repository by using Cursor CLI in GitHub Actions

Update documentation using Cursor CLI in GitHub Actions. Two approaches: full agent autonomy or deterministic workflow with agent-only file modifications.

<CodeGroup>
  ```yaml auto-update-docs.yml theme={null}
  name: Update Docs

on:
    pull_request:
      types: [opened, synchronize, reopened, ready_for_review]

permissions:
    contents: write
    pull-requests: write

jobs:
    auto-docs:
      if: ${{ !startsWith(github.head_ref, 'docs/') }}
      runs-on: ubuntu-latest
      steps:
        - name: Checkout repository
          uses: actions/checkout@v4
          with:
            fetch-depth: 0

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

- name: Update docs
          env:
            MODEL: gpt-5
            CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: docs
          run: |
            cursor-agent -p "You are operating in a GitHub Actions runner.

# Context:
            - Repo: ${{ github.repository }}
            - Owner: ${{ github.repository_owner }}
            - PR Number: ${{ github.event.pull_request.number }}
            - Base Ref: ${{ github.base_ref }}
            - Head Ref: ${{ github.head_ref }}
            - Docs Branch Prefix: ${{ env.BRANCH_PREFIX }}

# Goal:
            - Implement an end-to-end docs update flow driven by incremental changes to the original PR.

# Requirements:
            1) Determine what changed in the original PR and, if there have been multiple pushes, compute the incremental diffs since the last successful docs update.
            2) Update only the relevant docs based on those incremental changes.
            3) Maintain the persistent docs branch for this PR head using the Docs Branch Prefix from Context. Create it if missing, update it otherwise, and push changes to origin.
            4) You do NOT have permission to create PRs. Instead, post or update a single natural-language PR comment (1–2 sentences) that briefly explains the docs updates and includes an inline compare link to quick-create a PR

# Inputs and conventions:
            - Use `gh pr diff` and git history to detect changes and derive incremental ranges since the last docs update.
            - Do not attempt to create or edit PRs directly. Use the compare link format above.
            - Keep changes minimal and consistent with repo style. If no doc updates are necessary, make no changes and post no comment.

# Deliverables when updates occur:
            - Pushed commits to the persistent docs branch for this PR head.
            - A single natural-language PR comment on the original PR that includes the inline compare link above. Avoid posting duplicates; update a previous bot comment if present.
            " --force --model "$MODEL" --output-format=text
  ```

```yaml auto-update-docs-deterministic.yml theme={null}
  name: Update Docs

on:
    pull_request:
      types: [opened, synchronize, reopened, ready_for_review]

permissions:
    contents: write
    pull-requests: write

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

- name: Generate docs updates (no commit/push/comment)
          env:
            MODEL: gpt-5
            CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: docs
          run: |
            cursor-agent -p "You are operating in a GitHub Actions runner.

The GitHub CLI is available as `gh` and authenticated via `GH_TOKEN`. Git is available.

IMPORTANT: Do NOT create branches, commit, push, or post PR comments. Only modify files in the working directory as needed. A later workflow step is responsible for publishing changes and commenting on the PR.

# Goal:
            - Update repository documentation based on incremental changes introduced by this PR.

# Requirements:
            1) Determine what changed in the original PR (use `gh pr diff` and git history as needed). If an existing persistent docs branch `${{ env.BRANCH_PREFIX }}/${{ github.head_ref }}` exists, you may use it as a read-only reference point to understand prior updates.
            2) Update only the relevant docs based on those changes. Keep edits minimal and consistent with repo style.
            3) Do NOT commit, push, create branches, or post PR comments. Leave the working tree with updated files only; a later step will publish.

# Inputs and conventions:
            - Use `gh pr diff` and git history to detect changes and focus documentation edits accordingly.
            - If no doc updates are necessary, make no changes and produce no output.

# Deliverables when updates occur:
            - Modified documentation files in the working directory only (no commits/pushes/comments).
            " --force --model "$MODEL" --output-format=text

- name: Publish docs branch
          id: publish_docs
          env:
            BRANCH_PREFIX: docs
            HEAD_REF: ${{ github.head_ref }}
            PR_NUMBER: ${{ github.event.pull_request.number }}
          run: |
            echo "changes_published=false" >> "$GITHUB_OUTPUT"

DOCS_BRANCH="${BRANCH_PREFIX}/${HEAD_REF}"

# Ensure we are on a local branch that we can push
            git fetch origin --prune

# Create/switch to the persistent docs branch, keeping current working tree changes
            git checkout -B "$DOCS_BRANCH"

# Stage and detect changes
            git add -A
            if git diff --staged --quiet; then
              echo "No docs changes to publish. Skipping commit/push."
              exit 0
            fi

COMMIT_MSG="docs: update for PR #${PR_NUMBER} (${HEAD_REF} @ $(git rev-parse --short HEAD))"
            git commit -m "$COMMIT_MSG"
            git push --set-upstream origin "$DOCS_BRANCH"

echo "changes_published=true" >> "$GITHUB_OUTPUT"

- name: Post or update PR comment
          if: steps.publish_docs.outputs.changes_published == 'true'
          env:
            GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
            BRANCH_PREFIX: docs
            REPO: ${{ github.repository }}
            BASE_REF: ${{ github.base_ref }}
            HEAD_REF: ${{ github.head_ref }}
            PR_NUMBER: ${{ github.event.pull_request.number }}
          run: |

DOCS_BRANCH="${BRANCH_PREFIX}/${HEAD_REF}"
            COMPARE_URL="https://github.com/${REPO}/compare/${BASE_REF}...${DOCS_BRANCH}?quick_pull=1&title=docs%3A+updates+for+PR+%23${PR_NUMBER}"

COMMENT_FILE="${RUNNER_TEMP}/auto-docs-comment.md"
            {
              echo "Cursor updated docs branch: \`${DOCS_BRANCH}\`"
              echo "You can now [view the diff and quick-create a PR to merge these docs updates](${COMPARE_URL})."
              echo
              echo "_This comment will be updated on subsequent runs as the PR changes._"
              echo
              echo ""
            } > "$COMMENT_FILE"

# If editing the last bot comment fails (older gh), fall back to creating a new comment
            if gh pr comment "$PR_NUMBER" --body-file "$COMMENT_FILE" --edit-last; then
              echo "Updated existing PR comment."
            else
              gh pr comment "$PR_NUMBER" --body-file "$COMMENT_FILE"
              echo "Posted new PR comment."
            fi
  ```
</CodeGroup>

# GitHub Actions
Source: https://docs.cursor.com/en/cli/github-actions

Learn how to use Cursor CLI in GitHub Actions and other continuous integration systems

Use Cursor CLI in GitHub Actions and other CI/CD systems to automate development tasks.

## GitHub Actions integration

Basic setup:

```yaml  theme={null}
- name: Install Cursor CLI
  run: |
    curl https://cursor.com/install -fsS | bash
    echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Run Cursor Agent
  env:
    CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
  run: |
    cursor-agent -p "Your prompt here" --model gpt-5
```

## Cookbook examples

See our cookbook examples for practical workflows: [updating documentation](/en/cli/cookbook/update-docs) and [fixing CI issues](/en/cli/cookbook/fix-ci).

## Other CI systems

Use Cursor CLI in any CI/CD system with:

* **Shell script execution** (bash, zsh, etc.)
* **Environment variables** for API key configuration
* **Internet connectivity** to reach Cursor's API

## Autonomy levels

Choose your agent's autonomy level:

### Full autonomy approach

Give the agent complete control over git operations, API calls, and external interactions. Simpler setup, requires more trust.

**Example:** In our [Update Documentation](/en/cli/cookbook/update-docs) cookbook, the first workflow lets the agent:

* Analyze PR changes
* Create and manage git branches
* Commit and push changes
* Post comments on pull requests
* Handle all error scenarios

```yaml  theme={null}
- name: Update docs (full autonomy)
  run: |
    cursor-agent -p "You have full access to git, GitHub CLI, and PR operations. 
    Handle the entire docs update workflow including commits, pushes, and PR comments."
```

### Restricted autonomy approach

<Note>
  We recommend using this approach with **permission-based restrictions** for production CI workflows. This gives you the best of both worlds: the agent can intelligently handle complex analysis and file modifications while critical operations remain deterministic and auditable.
</Note>

Limit agent operations while handling critical steps in separate workflow steps. Better control and predictability.

**Example:** The second workflow in the same cookbook restricts the agent to only file modifications:

```yaml  theme={null}
- name: Generate docs updates (restricted)
  run: |
    cursor-agent -p "IMPORTANT: Do NOT create branches, commit, push, or post PR comments. 
    Only modify files in the working directory. A later workflow step handles publishing."

- name: Publish docs branch (deterministic)
  run: |
    # Deterministic git operations handled by CI
    git checkout -B "docs/${{ github.head_ref }}"
    git add -A
    git commit -m "docs: update for PR"
    git push origin "docs/${{ github.head_ref }}"

- name: Post PR comment (deterministic)  
  run: |
    # Deterministic PR commenting handled by CI
    gh pr comment ${{ github.event.pull_request.number }} --body "Docs updated"
```

### Permission-based restrictions

Use [permission configurations](/en/cli/reference/permissions) to enforce restrictions at the CLI level:

```json  theme={null}
{
  "permissions": {
    "allow": [
      "Read(**/*.md)",
      "Write(docs/**/*)",
      "Shell(grep)",
      "Shell(find)"
    ],
    "deny": [
      "Shell(git)",
      "Shell(gh)", 
      "Write(.env*)",
      "Write(package.json)"
    ]
  }
}
```

## Authentication

### Generate your API key

First, [generate an API key](/en/cli/reference/authentication#api-key-authentication) from your Cursor dashboard.

### Configure repository secrets

Store your Cursor API key securely in your repository:

1. Go to your GitHub repository
2. Click **Settings** → **Secrets and variables** → **Actions**
3. Click **New repository secret**
4. Name it `CURSOR_API_KEY`
5. Paste your API key as the value
6. Click **Add secret**

### Use in workflows

Set your `CURSOR_API_KEY` environment variable:

```yaml  theme={null}
env:
  CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
```

# Using Headless CLI
Source: https://docs.cursor.com/en/cli/headless

Learn how to write scripts using Cursor CLI for automated code analysis, generation, and modification

Use Cursor CLI in scripts and automation workflows for code analysis, generation, and refactoring tasks.

## How it works

Use [print mode](/en/cli/using#non-interactive-mode) (`-p, --print`) for non-interactive scripting and automation.

### File modification in scripts

Combine `--print` with `--force` to modify files in scripts:

```bash  theme={null}
# Enable file modifications in print mode
cursor-agent -p --force "Refactor this code to use modern ES6+ syntax"

# Without --force, changes are only proposed, not applied
cursor-agent -p "Add JSDoc comments to this file"  # Won't modify files

# Batch processing with actual file changes
find src/ -name "*.js" | while read file; do
  cursor-agent -p --force "Add comprehensive JSDoc comments to $file"
done
```

<Warning>
  The `--force` flag allows the agent to make direct file changes without confirmation
</Warning>

## Setup

See [Installation](/en/cli/installation) and [Authentication](/en/cli/reference/authentication) for complete setup details.

```bash  theme={null}
# Install Cursor CLI
curl https://cursor.com/install -fsS | bash

# Set API key for scripts  
export CURSOR_API_KEY=your_api_key_here
cursor-agent -p "Analyze this code"
```

## Example scripts

Use different output formats for different script needs. See [Output format](/en/cli/reference/output-format) for details.

### Searching the codebase

Use `--output-format text` for readable responses:

```bash  theme={null}
#!/bin/bash
# Simple codebase question

cursor-agent -p --output-format text "What does this codebase do?"
```

### Automated code review

Use `--output-format json` for structured analysis:

```bash  theme={null}
#!/bin/bash
# simple-code-review.sh - Basic code review script

echo "Starting code review..."

# Review recent changes
cursor-agent -p --force --output-format text \
  "Review the recent code changes and provide feedback on:
  - Code quality and readability  
  - Potential bugs or issues
  - Security considerations
  - Best practices compliance

Provide specific suggestions for improvement and write to review.txt"

if [ $? -eq 0 ]; then
  echo "✅ Code review completed successfully"
else
  echo "❌ Code review failed"
  exit 1
fi
```

### Real-time progress tracking

Use `--output-format stream-json` for real-time progress tracking:

```bash  theme={null}
#!/bin/bash
# stream-progress.sh - Track progress in real-time

echo "🚀 Starting stream processing..."

# Track progress in real-time
accumulated_text=""
tool_count=0
start_time=$(date +%s)

cursor-agent -p --force --output-format stream-json \
  "Analyze this project structure and create a summary report in analysis.txt" | \
  while IFS= read -r line; do
    
    type=$(echo "$line" | jq -r '.type // empty')
    subtype=$(echo "$line" | jq -r '.subtype // empty')
    
    case "$type" in
      "system")
        if [ "$subtype" = "init" ]; then
          model=$(echo "$line" | jq -r '.model // "unknown"')
          echo "🤖 Using model: $model"
        fi
        ;;
        
      "assistant")
        # Accumulate streaming text deltas
        content=$(echo "$line" | jq -r '.message.content[0].text // empty')
        accumulated_text="$accumulated_text$content"
        
        # Show live progress
        printf "\r📝 Generating: %d chars" ${#accumulated_text}
        ;;
        
      "tool_call")
        if [ "$subtype" = "started" ]; then
          tool_count=$((tool_count + 1))
          
          # Extract tool information
          if echo "$line" | jq -e '.tool_call.writeToolCall' > /dev/null 2>&1; then
            path=$(echo "$line" | jq -r '.tool_call.writeToolCall.args.path // "unknown"')
            echo -e "\n🔧 Tool #$tool_count: Creating $path"
          elif echo "$line" | jq -e '.tool_call.readToolCall' > /dev/null 2>&1; then
            path=$(echo "$line" | jq -r '.tool_call.readToolCall.args.path // "unknown"')
            echo -e "\n📖 Tool #$tool_count: Reading $path"
          fi
          
        elif [ "$subtype" = "completed" ]; then
          # Extract and show tool results
          if echo "$line" | jq -e '.tool_call.writeToolCall.result.success' > /dev/null 2>&1; then
            lines=$(echo "$line" | jq -r '.tool_call.writeToolCall.result.success.linesCreated // 0')
            size=$(echo "$line" | jq -r '.tool_call.writeToolCall.result.success.fileSize // 0')
            echo "   ✅ Created $lines lines ($size bytes)"
          elif echo "$line" | jq -e '.tool_call.readToolCall.result.success' > /dev/null 2>&1; then
            lines=$(echo "$line" | jq -r '.tool_call.readToolCall.result.success.totalLines // 0')
            echo "   ✅ Read $lines lines"
          fi
        fi
        ;;
        
      "result")
        duration=$(echo "$line" | jq -r '.duration_ms // 0')
        end_time=$(date +%s)
        total_time=$((end_time - start_time))
        
        echo -e "\n\n🎯 Completed in ${duration}ms (${total_time}s total)"
        echo "📊 Final stats: $tool_count tools, ${#accumulated_text} chars generated"
        ;;
    esac
  done
```

# Installation
Source: https://docs.cursor.com/en/cli/installation

Install and update Cursor CLI

## Installation

### macOS, Linux and Windows (WSL)

Install Cursor CLI with a single command:

```bash  theme={null}
curl https://cursor.com/install -fsS | bash
```

### Verification

After installation, verify that Cursor CLI is working correctly:

```bash  theme={null}
cursor-agent --version
```

## Post-installation setup

1. **Add \~/.local/bin to your PATH:**

For bash:

```bash  theme={null}
   echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
   source ~/.bashrc
   ```

For zsh:

```bash  theme={null}
   echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
   source ~/.zshrc
   ```

2. **Start using Cursor Agent:**
   ```bash  theme={null}
   cursor-agent
   ```

## Updates

Cursor CLI will try to auto-update by default to ensure you always have the latest version.

To manually update Cursor CLI to the latest version:

```bash  theme={null}
cursor-agent update
# or 
cursor-agent upgrade
```

Both commands will update Cursor Agent to the latest version.

# MCP
Source: https://docs.cursor.com/en/cli/mcp

Use MCP servers with cursor-agent to connect external tools and data sources

## Overview

The Cursor CLI supports [Model Context Protocol (MCP)](/en/context/mcp) servers, allowing you to connect external tools and data sources to `cursor-agent`. **MCP in the CLI uses the same configuration as the editor** - any MCP servers you've configured will work seamlessly with both.

<Card title="Learn about MCP" icon="link" href="/en/context/mcp">
  New to MCP? Read the complete guide on configuration, authentication, and available servers
</Card>

## CLI commands

Use the `cursor-agent mcp` command to manage MCP servers:

### List configured servers

View all configured MCP servers and their current status:

```bash  theme={null}
cursor-agent mcp list
```

This shows:

* Server names and identifiers
* Connection status (connected/disconnected)
* Configuration source (project or global)
* Transport method (stdio, HTTP, SSE)

### List available tools

View tools provided by a specific MCP server:

```bash  theme={null}
cursor-agent mcp list-tools <identifier>
```

This displays:

* Tool names and descriptions
* Required and optional parameters
* Parameter types and constraints

### Login to MCP server

Authenticate with an MCP server configured in your `mcp.json`:

```bash  theme={null}
cursor-agent mcp login <identifier>
```

### Disable MCP server

Remove an MCP server from the local approved list:

```bash  theme={null}
cursor-agent mcp disable <identifier>
```

## Using MCP with Agent

Once you have MCP servers configured (see the [main MCP guide](/en/context/mcp) for setup), `cursor-agent` automatically discovers and uses available tools when relevant to your requests.

```bash  theme={null}
# Check what MCP servers are available
cursor-agent mcp list

# See what tools a specific server provides
cursor-agent mcp list-tools playwright

# Use cursor-agent - it automatically uses MCP tools when helpful
cursor-agent --prompt "Navigate to google.com and take a screenshot of the search page"
```

The CLI follows the same configuration precedence as the editor (project → global → nested), automatically discovering configurations from parent directories.

## Related

<CardGroup cols={2}>
  <Card title="MCP Overview" icon="link" href="/en/context/mcp">
    Complete MCP guide: setup, configuration, and authentication
  </Card>

<Card title="Available MCP Tools" icon="table" href="/en/tools">
    Browse pre-built MCP servers you can use
  </Card>
</CardGroup>

# Cursor CLI
Source: https://docs.cursor.com/en/cli/overview

Get started with Cursor CLI to code in your terminal

Cursor CLI lets you interact with AI agents directly from your terminal to write, review, and modify code. Whether you prefer an interactive terminal interface or print automation for scripts and CI pipelines, the CLI provides powerful coding assistance right where you work.

```bash  theme={null}
# Install
curl https://cursor.com/install -fsS | bash

# Run interactive session
cursor-agent
```

<Info>
  Cursor CLI is currently in beta, we'd love your feedback on it!
</Info>

### Interactive mode

Start a conversational session with the agent to describe your goals, review proposed changes, and approve commands:

```bash  theme={null}
# Start interactive session
cursor-agent

# Start with initial prompt
cursor-agent "refactor the auth module to use JWT tokens"
```

### Non-interactive mode

Use print mode for non-interactive scenarios like scripts, CI pipelines, or automation:

```bash  theme={null}
# Run with specific prompt and model
cursor-agent -p "find and fix performance issues" --model "gpt-5"

# Use with git changes included for review
cursor-agent -p "review these changes for security issues" --output-format text
```

### Sessions

Resume previous conversations to maintain context across multiple interactions:

```bash  theme={null}
# List all previous chats
cursor-agent ls

# Resume latest conversation  
cursor-agent resume

# Resume specific conversation
cursor-agent --resume="chat-id-here"
```

# Authentication
Source: https://docs.cursor.com/en/cli/reference/authentication

Authenticate Cursor CLI using browser flow or API keys

Cursor CLI supports two authentication methods: browser-based login (recommended) and API keys.

## Browser authentication (recommended)

Use the browser flow for the easiest authentication experience:

```bash  theme={null}
# Log in using browser flow
cursor-agent login

# Check authentication status
cursor-agent status

# Log out and clear stored authentication
cursor-agent logout
```

The login command will open your default browser and prompt you to authenticate with your Cursor account. Once completed, your credentials are securely stored locally.

## API key authentication

For automation, scripts, or CI/CD environments, use API key authentication:

### Step 1: Generate an API key

Generate an API key in your Cursor dashboard under Integrations > User API Keys.

### Step 2: Set the API key

You can provide the API key in two ways:

**Option 1: Environment variable (recommended)**

```bash  theme={null}
export CURSOR_API_KEY=your_api_key_here
cursor-agent "implement user authentication"
```

**Option 2: Command line flag**

```bash  theme={null}
cursor-agent --api-key your_api_key_here "implement user authentication"
```

## Authentication status

Check your current authentication status:

```bash  theme={null}
cursor-agent status
```

This command will display:

* Whether you're authenticated
* Your account information
* Current endpoint configuration

## Troubleshooting

* **"Not authenticated" errors:** Run `cursor-agent login` or ensure your API key is correctly set
* **SSL certificate errors:** Use the `--insecure` flag for development environments
* **Endpoint issues:** Use the `--endpoint` flag to specify a custom API endpoint

# Configuration
Source: https://docs.cursor.com/en/cli/reference/configuration

Agent CLI configuration reference for cli-config.json

Configure the Agent CLI using the `cli-config.json` file.

## File location

<Note>Only permissions can be configured at the project level. All other CLI settings must be set globally.</Note>

Override with environment variables:

* **`CURSOR_CONFIG_DIR`**: custom directory path
* **`XDG_CONFIG_HOME`** (Linux/BSD): uses `$XDG_CONFIG_HOME/cursor/cli-config.json`

## Schema

### Required fields

### Optional fields

## Examples

### Minimal config

```json  theme={null}
{
  "version": 1,
  "editor": { "vimMode": false },
  "permissions": { "allow": ["Shell(ls)"], "deny": [] }
}
```

### Enable Vim mode

```json  theme={null}
{
  "version": 1,
  "editor": { "vimMode": true },
  "permissions": { "allow": ["Shell(ls)"], "deny": [] }
}
```

### Configure permissions

```json  theme={null}
{
  "version": 1,
  "editor": { "vimMode": false },
  "permissions": {
    "allow": ["Shell(ls)", "Shell(echo)"],
    "deny": ["Shell(rm)"]
  }
}
```

See [Permissions](/en/cli/reference/permissions) for available permission types and examples.

## Troubleshooting

**Config errors**: Move the file aside and restart:

```bash  theme={null}
mv ~/.cursor/cli-config.json ~/.cursor/cli-config.json.bad
```

**Changes don't persist**: Ensure valid JSON and write permissions. Some fields are CLI-managed and may be overwritten.

## Notes

* Pure JSON format (no comments)
* CLI performs self-repair for missing fields
* Corrupted files are backed up as `.bad` and recreated
* Permission entries are exact strings (see [Permissions](/en/cli/reference/permissions) for details)

# Output format
Source: https://docs.cursor.com/en/cli/reference/output-format

Output schema for text, JSON and stream-JSON formats

The Cursor Agent CLI provides multiple output formats with the `--output-format` option when combined with `--print`. These formats include structured formats for programmatic use (`json`, `stream-json`) and a simplified text format for human-readable progress tracking.

<Note>
  The default `--output-format` is `stream-json`. This option is only valid when printing (`--print`) or when print mode is inferred (non-TTY stdout or piped stdin).
</Note>

## JSON format

The `json` output format emits a single JSON object (followed by a newline) when the run completes successfully. Deltas and tool events are not emitted; text is aggregated into the final result.

On failure, the process exits with a non-zero code and writes an error message to stderr. No well-formed JSON object is emitted in failure cases.

### Success response

When successful, the CLI outputs a JSON object with the following structure:

```json  theme={null}
{
  "type": "result",
  "subtype": "success",
  "is_error": false,
  "duration_ms": 1234,
  "duration_api_ms": 1234,
  "result": "<full assistant text>",
  "session_id": "<uuid>",
  "request_id": "<optional request id>"
}
```

<div class="full-width-table">
  | Field             | Description                                                         |
  | ----------------- | ------------------------------------------------------------------- |
  | `type`            | Always `"result"` for terminal results                              |
  | `subtype`         | Always `"success"` for successful completions                       |
  | `is_error`        | Always `false` for successful responses                             |
  | `duration_ms`     | Total execution time in milliseconds                                |
  | `duration_api_ms` | API request time in milliseconds (currently equal to `duration_ms`) |
  | `result`          | Complete assistant response text (concatenation of all text deltas) |
  | `session_id`      | Unique session identifier                                           |
  | `request_id`      | Optional request identifier (may be omitted)                        |
</div>

## Stream JSON format

The `stream-json` output format emits newline-delimited JSON (NDJSON). Each line contains a single JSON object representing a real-time event during execution.

The stream ends with a terminal `result` event on success. On failure, the process exits with a non-zero code and the stream may end early without a terminal event; an error message is written to stderr.

### Event types

#### System initialization

Emitted once at the beginning of each session:

```json  theme={null}
{
  "type": "system",
  "subtype": "init",
  "apiKeySource": "env|flag|login",
  "cwd": "/absolute/path",
  "session_id": "<uuid>",
  "model": "<model display name>",
  "permissionMode": "default"
}
```

<Note>
  Future fields like `tools` and `mcp_servers` may be added to this event.
</Note>

#### User message

Contains the user's input prompt:

```json  theme={null}
{
  "type": "user",
  "message": {
    "role": "user",
    "content": [{ "type": "text", "text": "<prompt>" }]
  },
  "session_id": "<uuid>"
}
```

#### Assistant text delta

Emitted multiple times as the assistant generates its response. These events contain incremental text chunks:

```json  theme={null}
{
  "type": "assistant",
  "message": {
    "role": "assistant",
    "content": [{ "type": "text", "text": "<delta chunk>" }]
  },
  "session_id": "<uuid>"
}
```

<Note>
  Concatenate all `message.content[].text` values in order to reconstruct the complete assistant response.
</Note>

#### Tool call events

Tool calls are tracked with start and completion events:

**Tool call started:**

```json  theme={null}
{
  "type": "tool_call",
  "subtype": "started",
  "call_id": "<string id>",
  "tool_call": {
    "readToolCall": {
      "args": { "path": "file.txt" }
    }
  },
  "session_id": "<uuid>"
}
```

**Tool call completed:**

```json  theme={null}
{
  "type": "tool_call",
  "subtype": "completed",
  "call_id": "<string id>",
  "tool_call": {
    "readToolCall": {
      "args": { "path": "file.txt" },
      "result": {
        "success": {
          "content": "file contents...",
          "isEmpty": false,
          "exceededLimit": false,
          "totalLines": 54,
          "totalChars": 1254
        }
      }
    }
  },
  "session_id": "<uuid>"
}
```

#### Tool call types

**Read file tool:**

* **Started**: `tool_call.readToolCall.args` contains `{ "path": "file.txt" }`
* **Completed**: `tool_call.readToolCall.result.success` contains file metadata and content

**Write file tool:**

* **Started**: `tool_call.writeToolCall.args` contains `{ "path": "file.txt", "fileText": "content...", "toolCallId": "id" }`
* **Completed**: `tool_call.writeToolCall.result.success` contains `{ "path": "/absolute/path", "linesCreated": 19, "fileSize": 942 }`

**Other tools:**

* May use `tool_call.function` structure with `{ "name": "tool_name", "arguments": "..." }`

#### Terminal result

The final event emitted on successful completion:

```json  theme={null}
{
  "type": "result",
  "subtype": "success",
  "duration_ms": 1234,
  "duration_api_ms": 1234,
  "is_error": false,
  "result": "<full assistant text>",
  "session_id": "<uuid>",
  "request_id": "<optional request id>"
}
```

### Example sequence

Here's a representative NDJSON sequence showing the typical flow of events:

```json  theme={null}
{"type":"system","subtype":"init","apiKeySource":"login","cwd":"/Users/user/project","session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff","model":"Claude 4 Sonnet","permissionMode":"default"}
{"type":"user","message":{"role":"user","content":[{"type":"text","text":"Read README.md and create a summary"}]},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"I'll "}]},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"read the README.md file"}]},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"tool_call","subtype":"started","call_id":"toolu_vrtx_01NnjaR886UcE8whekg2MGJd","tool_call":{"readToolCall":{"args":{"path":"README.md"}}},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"tool_call","subtype":"completed","call_id":"toolu_vrtx_01NnjaR886UcE8whekg2MGJd","tool_call":{"readToolCall":{"args":{"path":"README.md"},"result":{"success":{"content":"# Project\n\nThis is a sample project...","isEmpty":false,"exceededLimit":false,"totalLines":54,"totalChars":1254}}}},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":" and create a summary"}]},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"tool_call","subtype":"started","call_id":"toolu_vrtx_01Q3VHVnWFSKygaRPT7WDxrv","tool_call":{"writeToolCall":{"args":{"path":"summary.txt","fileText":"# README Summary\n\nThis project contains...","toolCallId":"toolu_vrtx_01Q3VHVnWFSKygaRPT7WDxrv"}}},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"tool_call","subtype":"completed","call_id":"toolu_vrtx_01Q3VHVnWFSKygaRPT7WDxrv","tool_call":{"writeToolCall":{"args":{"path":"summary.txt","fileText":"# README Summary\n\nThis project contains...","toolCallId":"toolu_vrtx_01Q3VHVnWFSKygaRPT7WDxrv"},"result":{"success":{"path":"/Users/user/project/summary.txt","linesCreated":19,"fileSize":942}}}},"session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff"}
{"type":"result","subtype":"success","duration_ms":5234,"duration_api_ms":5234,"is_error":false,"result":"I'll read the README.md file and create a summary","session_id":"c6b62c6f-7ead-4fd6-9922-e952131177ff","request_id":"10e11780-df2f-45dc-a1ff-4540af32e9c0"}
```

## Text format

The `text` output format provides a simplified, human-readable stream of agent actions. Instead of detailed JSON events, it outputs concise text descriptions of what the agent is doing in real-time.

This format is useful for monitoring agent progress without the overhead of parsing structured data, making it ideal for logging, debugging, or simple progress tracking.

### Example output

```
Read file
Edited file
Ran terminal command
Created new file
```

Each action appears on a new line as the agent performs it, providing immediate feedback on the agent's progress through the task.

## Implementation notes

* Each event is emitted as a single line terminated by `\n`
* `thinking` events are suppressed in print mode and will not appear in either output format
* Field additions may occur over time in a backward-compatible way (consumers should ignore unknown fields)
* The stream format provides real-time updates, while the JSON format waits for completion before outputting results
* Concatenate all `assistant` message deltas to reconstruct the complete response
* Tool call IDs can be used to correlate start/completion events
* Session IDs remain consistent throughout a single agent execution

# Parameters
Source: https://docs.cursor.com/en/cli/reference/parameters

Complete command reference for Cursor Agent CLI

## Global options

Global options can be used with any command:

<div class="full-width-table">
  | Option                     | Description                                                                                                         |
  | -------------------------- | ------------------------------------------------------------------------------------------------------------------- |
  | `-v, --version`            | Output the version number                                                                                           |
  | `-a, --api-key <key>`      | API key for authentication (can also use `CURSOR_API_KEY` env var)                                                  |
  | `-p, --print`              | Print responses to console (for scripts or non-interactive use). Has access to all tools, including write and bash. |
  | `--output-format <format>` | Output format (only works with `--print`): `text`, `json`, or `stream-json` (default: `stream-json`)                |
  | `-b, --background`         | Start in background mode (open composer picker on launch)                                                           |
  | `--fullscreen`             | Enable fullscreen mode                                                                                              |
  | `--resume [chatId]`        | Resume a chat session                                                                                               |
  | `-m, --model <model>`      | Model to use                                                                                                        |
  | `-f, --force`              | Force allow commands unless explicitly denied                                                                       |
  | `-h, --help`               | Display help for command                                                                                            |
</div>

## Commands

<div class="full-width-table">
  | Command           | Description                               | Usage                                           |
  | ----------------- | ----------------------------------------- | ----------------------------------------------- |
  | `login`           | Authenticate with Cursor                  | `cursor-agent login`                            |
  | `logout`          | Sign out and clear stored authentication  | `cursor-agent logout`                           |
  | `status`          | Check authentication status               | `cursor-agent status`                           |
  | `mcp`             | Manage MCP servers                        | `cursor-agent mcp`                              |
  | `update\|upgrade` | Update Cursor Agent to the latest version | `cursor-agent update` or `cursor-agent upgrade` |
  | `ls`              | Resume a chat session                     | `cursor-agent ls`                               |
  | `resume`          | Resume the latest chat session            | `cursor-agent resume`                           |
  | `help [command]`  | Display help for command                  | `cursor-agent help [command]`                   |
</div>

<Note>
  When no command is specified, Cursor Agent starts in interactive chat mode by default.
</Note>

## MCP

Manage MCP servers configured for Cursor Agent.

<div class="full-width-table">
  | Subcommand                | Description                                                      | Usage                                      |
  | ------------------------- | ---------------------------------------------------------------- | ------------------------------------------ |
  | `login <identifier>`      | Authenticate with an MCP server configured in `.cursor/mcp.json` | `cursor-agent mcp login <identifier>`      |
  | `list`                    | List configured MCP servers and their status                     | `cursor-agent mcp list`                    |
  | `list-tools <identifier>` | List available tools and their argument names for a specific MCP | `cursor-agent mcp list-tools <identifier>` |
</div>

All MCP commands support `-h, --help` for command-specific help.

## Arguments

When starting in chat mode (default behavior), you can provide an initial prompt:

**Arguments:**

* `prompt` — Initial prompt for the agent

## Getting help

All commands support the global `-h, --help` option to display command-specific help.

# Permissions
Source: https://docs.cursor.com/en/cli/reference/permissions

Permission types for controlling agent access to files and commands

Configure what the agent is allowed to do using permission tokens in your CLI configuration. Permissions are set in `~/.cursor/cli-config.json` (global) or `<project>/.cursor/cli.json` (project-specific).

## Permission types

### Shell commands

**Format:** `Shell(commandBase)`

Controls access to shell commands. The `commandBase` is the first token in the command line.

<div class="full-width-table">
  | Example      | Description                                        |
  | ------------ | -------------------------------------------------- |
  | `Shell(ls)`  | Allow running `ls` commands                        |
  | `Shell(git)` | Allow any `git` subcommand                         |
  | `Shell(npm)` | Allow npm package manager commands                 |
  | `Shell(rm)`  | Deny destructive file removal (commonly in `deny`) |
</div>

### File reads

**Format:** `Read(pathOrGlob)`

Controls read access to files and directories. Supports glob patterns.

<div class="full-width-table">
  | Example             | Description                             |
  | ------------------- | --------------------------------------- |
  | `Read(src/**/*.ts)` | Allow reading TypeScript files in `src` |
  | `Read(**/*.md)`     | Allow reading markdown files anywhere   |
  | `Read(.env*)`       | Deny reading environment files          |
  | `Read(/etc/passwd)` | Deny reading system files               |
</div>

### File writes

**Format:** `Write(pathOrGlob)`

Controls write access to files and directories. Supports glob patterns. When using in print mode, `--force` is required to write files.

<div class="full-width-table">
  | Example               | Description                           |
  | --------------------- | ------------------------------------- |
  | `Write(src/**)`       | Allow writing to any file under `src` |
  | `Write(package.json)` | Allow modifying package.json          |
  | `Write(**/*.key)`     | Deny writing private key files        |
  | `Write(**/.env*)`     | Deny writing environment files        |
</div>

## Configuration

Add permissions to the `permissions` object in your CLI configuration file:

```json  theme={null}
{
  "permissions": {
    "allow": [
      "Shell(ls)",
      "Shell(git)", 
      "Read(src/**/*.ts)",
      "Write(package.json)"
    ],
    "deny": [
      "Shell(rm)",
      "Read(.env*)",
      "Write(**/*.key)"
    ]
  }
}
```

## Pattern matching

* Glob patterns use `**`, `*`, and `?` wildcards
* Relative paths are scoped to the current workspace
* Absolute paths can target files outside the project
* Deny rules take precedence over allow rules

# Slash commands
Source: https://docs.cursor.com/en/cli/reference/slash-commands

Quick actions available within Cursor CLI sessions

<div class="full-width-table">
  | Command               | Description                                         |
  | --------------------- | --------------------------------------------------- |
  | `/model <model>`      | Set or list models                                  |
  | `/auto-run [state]`   | Toggle auto-run (default) or set \[on\|off\|status] |
  | `/new-chat`           | Start a new chat session                            |
  | `/vim`                | Toggle Vim keys                                     |
  | `/help [command]`     | Show help (/help \[cmd])                            |
  | `/feedback <message>` | Share feedback with the team                        |
  | `/resume <chat>`      | Resume a previous chat by folder name               |
  | `/copy-req-id`        | Copy last request ID                                |
  | `/logout`             | Sign out from Cursor                                |
  | `/quit`               | Exit                                                |
</div>

# Shell Mode
Source: https://docs.cursor.com/en/cli/shell-mode

Run shell commands directly from the CLI without leaving your conversation

Shell Mode runs shell commands directly from the CLI without leaving your conversation. Use it for quick, non-interactive commands with safety checks and output displayed in the conversation.

## Command execution

Commands run in your login shell (`$SHELL`) with the CLI's working directory and environment. Chain commands to run in other directories:

```bash  theme={null}
cd subdir && npm test
```

## Output

<product_visual type="screenshot">
  Command output showing header with exit code, stdout/stderr display, and truncation controls
</product_visual>

Large outputs are truncated automatically and long-running processes timeout to maintain performance.

## Limitations

* Commands timeout after 30 seconds
* Long-running processes, servers, and interactive prompts are not supported
* Use short, non-interactive commands for best results

## Permissions

Commands are checked against your permissions and team settings before execution. See [Permissions](/en/cli/reference/permissions) for detailed configuration.

<product_visual type="screenshot">
  Decision banner showing approval options: Run, Reject/Propose, Add to allowlist, and Auto-run
</product_visual>

Admin policies may block certain commands, and commands with redirection cannot be allowlisted inline.

## Usage guidelines

Shell Mode works well for status checks, quick builds, file operations, and environment inspection.

Avoid long-running servers, interactive applications, and commands requiring input.

Each command runs independently - use `cd <dir> && ...` to run commands in other directories.

## Troubleshooting

* If a command hangs, cancel with <Kbd>Ctrl+C</Kbd> and add non-interactive flags
* When prompted for permissions, approve once or add to allowlist with <Kbd>Tab</Kbd>
* For truncated output, use <Kbd>Ctrl+O</Kbd> to expand
* To run in different directories, use `cd <dir> && ...` since changes don't persist
* Shell Mode supports zsh and bash from your `$SHELL` variable

## FAQ

<AccordionGroup>
  <Accordion title="Does `cd` persist across runs?">
    No. Each command runs independently. Use `cd <dir> && ...` to run commands in different directories.
  </Accordion>

<Accordion title="Can I change the timeout?">
    No. Commands are limited to 30 seconds and this is not configurable.
  </Accordion>

<Accordion title="Where are permissions configured?">
    Permissions are managed by CLI and team configuration. Use the decision banner to add commands to allowlists.
  </Accordion>

<Accordion title="How do I exit Shell Mode?">
    Press <Kbd>Escape</Kbd> when the input is empty, <Kbd>Backspace</Kbd>/<Kbd>Delete</Kbd> on empty input, or <Kbd>Ctrl+C</Kbd> to clear and exit.
  </Accordion>
</AccordionGroup>

# Using Agent in CLI
Source: https://docs.cursor.com/en/cli/using

Prompt, review, and iterate effectively with Cursor CLI

## Prompting

Stating intent clearly is recommended for the best results. For example, you can use the prompt "do not write any code" to ensure that the agent won't edit any files. This is generally helpful when planning tasks before implementing them.

Agent currently has tools for file operations, searching, and running shell commands. More tools are being added, similar to the IDE agent.

## MCP

Agent supports [MCP (Model Context Protocol)](/en/tools/mcp) for extended functionality and integrations. The CLI will automatically detect and respect your `mcp.json` configuration file, enabling the same MCP servers and tools that you've configured for the IDE.

## Rules

The CLI agent supports the same [rules system](/en/context/rules) as the IDE. You can create rules in the `.cursor/rules` directory to provide context and guidance to the agent. These rules will be automatically loaded and applied based on their configuration, allowing you to customize the agent's behavior for different parts of your project or specific file types.

<Note>
  The CLI also reads `AGENTS.md` and `CLAUDE.md` at the project root (if present) and applies them as rules alongside `.cursor/rules`.
</Note>

## Working with Agent

### Navigation

Previous messages can be accessed using arrow up (<Kbd>ArrowUp</Kbd>) where you can cycle through them.

### Review

Review changes with <Kbd>Cmd+R</Kbd>. Press <Kbd>i</Kbd> to add follow-up instructions. Use <Kbd>ArrowUp</Kbd>/<Kbd>ArrowDown</Kbd> to scroll, and <Kbd>ArrowLeft</Kbd>/<Kbd>ArrowRight</Kbd> to switch files.

### Selecting context

Select files and folders to include in context with <Kbd>@</Kbd>. Free up space in the context window by running `/compress`. See [Summarization](/en/agent/chat/summarization) for details.

## History

Continue from an existing thread with `--resume [thread id]` to load prior context.

To resume the most recent conversation, use `cursor-agent resume`.

You can also run `cursor-agent ls` to see a list of previous conversations.

## Command approval

Before running terminal commands, CLI will ask you to approve (<Kbd>y</Kbd>) or reject (<Kbd>n</Kbd>) execution.

## Non-interactive mode

Use `-p` or `--print` to run Agent in non-interactive mode. This will print the response to the console.

With non-interactive mode, you can invoke Agent in a non-interactive way. This allows you to integrate it in scripts, CI pipelines, etc.

You can combine this with `--output-format` to control how the output is formatted. For example, use `--output-format json` for structured output that's easier to parse in scripts, or `--output-format text` for plain text output.

<Note>
  Cursor has full write access in non-interactive mode.
</Note>

# Keyboard Shortcuts
Source: https://docs.cursor.com/en/configuration/kbd

Keyboard shortcuts and keybindings in Cursor

Overview of keyboard shortcuts in Cursor. See all keyboard shortcuts by pressing <Kbd>Cmd R</Kbd> then <Kbd>Cmd S</Kbd> or by opening command palette <Kbd>Cmd Shift P</Kbd> and searching for `Keyboard Shortcuts`.

Learn more about Keyboard Shortcuts in Cursor with [Key Bindings for VS Code](https://code.visualstudio.com/docs/getstarted/keybindings) as a baseline for Cursor's keybindings.

All Cursor keybindings, including Cursor-specific features, can be remapped in Keyboard Shortcuts settings.

## General

<div className="full-width-table equal-table-columns">
  | Shortcut               | Action                                  |
  | ---------------------- | --------------------------------------- |
  | <Kbd>Cmd I</Kbd>       | Toggle Sidepanel (unless bound to mode) |
  | <Kbd>Cmd L</Kbd>       | Toggle Sidepanel (unless bound to mode) |
  | <Kbd>Cmd E</Kbd>       | Background Agent control panel          |
  | <Kbd>Cmd .</Kbd>       | Mode Menu                               |
  | <Kbd>Cmd /</Kbd>       | Loop between AI models                  |
  | <Kbd>Cmd Shift J</Kbd> | Cursor settings                         |
  | <Kbd>Cmd ,</Kbd>       | General settings                        |
  | <Kbd>Cmd Shift P</Kbd> | Command palette                         |
</div>

## Chat

Shortcuts for the chat input box.

<div className="full-width-table equal-table-columns">
  | Shortcut                                             | Action                       |
  | ---------------------------------------------------- | ---------------------------- |
  | <Kbd>Return</Kbd>                                    | Nudge (default)              |
  | <Kbd>Ctrl Return</Kbd>                               | Queue message                |
  | <Kbd>Cmd Return</Kbd> when typing                    | Force send message           |
  | <Kbd>Cmd Shift Backspace</Kbd>                       | Cancel generation            |
  | <Kbd>Cmd Shift L</Kbd> with code selected            | Add selected code as context |
  | <Kbd>Cmd V</Kbd> with code or log in clipboard       | Add clipboard as context     |
  | <Kbd>Cmd Shift V</Kbd> with code or log in clipboard | Add clipboard to input box   |
  | <Kbd>Cmd Return</Kbd> with suggested changes         | Accept all changes           |
  | <Kbd>Cmd Backspace</Kbd>                             | Reject all changes           |
  | <Kbd>Tab</Kbd>                                       | Cycle to next message        |
  | <Kbd>Shift Tab</Kbd>                                 | Cycle to previous message    |
  | <Kbd>Cmd Opt /</Kbd>                                 | Model toggle                 |
  | <Kbd>Cmd N</Kbd> / <Kbd>Cmd R</Kbd>                  | New chat                     |
  | <Kbd>Cmd T</Kbd>                                     | New chat tab                 |
  | <Kbd>Cmd \[</Kbd>                                    | Previous chat                |
  | <Kbd>Cmd ]</Kbd>                                     | Next chat                    |
  | <Kbd>Cmd W</Kbd>                                     | Close chat                   |
  | <Kbd>Escape</Kbd>                                    | Unfocus field                |
</div>

## Inline Edit

<div className="full-width-table equal-table-columns">
  | Shortcut                       | Action             |
  | ------------------------------ | ------------------ |
  | <Kbd>Cmd K</Kbd>               | Open               |
  | <Kbd>Cmd Shift K</Kbd>         | Toggle input focus |
  | <Kbd>Return</Kbd>              | Submit             |
  | <Kbd>Cmd Shift Backspace</Kbd> | Cancel             |
  | <Kbd>Opt Return</Kbd>          | Ask quick question |
</div>

## Code Selection & Context

<div className="full-width-table equal-table-columns">
  | Shortcut                                              | Action                               |
  | ----------------------------------------------------- | ------------------------------------ |
  | <Kbd>@</Kbd>                                          | [@-symbols](/en/context/@-symbols/)  |
  | <Kbd>#</Kbd>                                          | Files                                |
  | <Kbd>/</Kbd>                                          | Shortcut Commands                    |
  | <Kbd>Cmd Shift L</Kbd>                                | Add selection to Chat                |
  | <Kbd>Cmd Shift K</Kbd>                                | Add selection to Edit                |
  | <Kbd>Cmd L</Kbd>                                      | Add selection to new chat            |
  | <Kbd>Cmd M</Kbd>                                      | Toggle file reading strategies       |
  | <Kbd>Cmd →</Kbd>                                      | Accept next word of suggestion       |
  | <Kbd>Cmd Return</Kbd>                                 | Search codebase in chat              |
  | Select code, <Kbd>Cmd C</Kbd>, <Kbd>Cmd V</Kbd>       | Add copied reference code as context |
  | Select code, <Kbd>Cmd C</Kbd>, <Kbd>Cmd Shift V</Kbd> | Add copied code as text context      |
</div>

## Tab

<div className="full-width-table equal-table-columns">
  | Shortcut         | Action            |
  | ---------------- | ----------------- |
  | <Kbd>Tab</Kbd>   | Accept suggestion |
  | <Kbd>Cmd →</Kbd> | Accept next word  |
</div>

## Terminal

<div className="full-width-table equal-table-columns">
  | Shortcut              | Action                   |
  | --------------------- | ------------------------ |
  | <Kbd>Cmd K</Kbd>      | Open terminal prompt bar |
  | <Kbd>Cmd Return</Kbd> | Run generated command    |
  | <Kbd>Escape</Kbd>     | Accept command           |
</div>

# Shell Commands
Source: https://docs.cursor.com/en/configuration/shell

Install and use Cursor shell commands

Cursor provides command-line tools to open files and folders from your terminal. Install both the `cursor` and `code` commands to integrate Cursor with your development workflow.

## Installing CLI commands

Install the CLI commands through the Command Palette:

1. Open the Command Palette (Cmd/Ctrl + P)
2. Type "Install" to filter installation commands
3. Select and run `Install 'cursor' to shell`
4. Repeat and select `Install 'code' to shell`

<product_visual type="screenshot">
  Command Palette showing CLI installation options
</product_visual>

## Using the CLI commands

After installation, use either command to open files or folders in Cursor:

```bash  theme={null}
# Using the cursor command
cursor path/to/file.js
cursor path/to/folder/

# Using the code command (VS Code compatible)
code path/to/file.js
code path/to/folder/
```

## Command options

Both commands support these options:

* Open a file: `cursor file.js`
* Open a folder: `cursor ./my-project`
* Open multiple items: `cursor file1.js file2.js folder1/`
* Open in a new window: `cursor -n` or `cursor --new-window`
* Wait for the window to close: `cursor -w` or `cursor --wait`

## FAQ

<AccordionGroup>
  <Accordion title="What's the difference between cursor and code commands?">
    They're identical. The `code` command is provided for VS Code compatibility.
  </Accordion>

<Accordion title="Do I need to install both commands?">
    No, install either or both based on preference.
  </Accordion>

<Accordion title="Where are the commands installed?">
    Commands are installed in your system's default shell configuration file (e.g., `.bashrc`, `.zshrc`, or `.config/fish/config.fish`).
  </Accordion>
</AccordionGroup>

# Themes
Source: https://docs.cursor.com/en/configuration/themes

Customize the appearance of Cursor

Cursor supports both light and dark themes for your coding environment. Cursor inherits VS Code's theming capabilities - use any VS Code theme, create custom themes, and install theme extensions from the marketplace.

## Changing theme

1. Open the Command Palette (Cmd/Ctrl + P)
2. Type "theme" to filter commands
3. Select "Preferences: Color Theme"
4. Choose a theme

## FAQ

<AccordionGroup>
  <Accordion title="Can I use my VS Code themes in Cursor?">
    Yes! Cursor is compatible with VS Code themes. Install any VS Code marketplace theme or copy custom theme files.
  </Accordion>

<Accordion title="How do I create a custom theme?">
    Create custom themes like in VS Code. Use "Developer: Generate Color Theme From Current Settings" to start from current settings, or follow the VS Code theme authoring guide.
  </Accordion>
</AccordionGroup>

# @Code
Source: https://docs.cursor.com/en/context/@-symbols/@-code

Reference specific code snippets in Cursor using @Code

Reference specific code sections using the `@Code` symbol. This provides more granular control than [`@Files & Folders`](/en/context/@-symbols/@-files-and-folders), letting you select precise code snippets instead of entire files.

# @Cursor Rules
Source: https://docs.cursor.com/en/context/@-symbols/@-cursor-rules

Apply project-specific rules and guidelines

The `@Cursor Rules` symbol provides access to [project rules](/en/context/rules) and guidelines you've set up, letting you explicitly apply them to your context.

# @Files & Folders
Source: https://docs.cursor.com/en/context/@-symbols/@-files-and-folders

Reference files and folders as context in Chat and Inline Edit

## Files

Reference entire files in Chat and Inline Edit by selecting `@Files & Folders` followed by the filename to search. You can also drag files from the sidebar directly into Agent to add as context.

## Folders

When referencing folders using `@Folders`, Cursor provides the folder path and overview of its contents to help the AI understand what's available.

<Tip>
  After selecting a folder, type `/` to navigate deeper and see all subfolders.
</Tip>

### Full folder content

Enable **Full Folder Content** in settings. When enabled, Cursor attempts to include all files from the folder in the context.

For large folders exceeding the context window, an outline view appears with a tooltip showing how many files were included while Cursor manages the available context space.

<Note>
  Using full folder content with [Max mode enabled](/en/context/max-mode)
  significantly increases request costs as more input tokens are consumed.
</Note>

## Context management

Large files and folders are automatically condensed to fit within context limits. See [file & folder condensation](/en/agent/chats/summarization#file--folder-condensation) for details.

# @Git
Source: https://docs.cursor.com/en/context/@-symbols/@-git

Reference Git changes and branch differences

* `@Commit`: Reference current working state changes compared to the last commit. Shows all modified, added, and deleted files not yet committed.
* `@Branch`: Compare your current branch's changes with the main branch. Shows all commits and changes in your branch but not in main.

# @Link
Source: https://docs.cursor.com/en/context/@-symbols/@-link

Include web content by pasting URLs

When you paste a URL in Chat, Cursor automatically tags it as an `@Link` and fetches the content to use as context. This includes support for PDF documents - Cursor extracts and parses text content from any publicly accessible PDF URL.

## Unlink

To use a URL as plain text without fetching its content:

* Click the tagged link and select `Unlink`
* Or paste while holding `Shift` to prevent automatic tagging

# @Linter Errors
Source: https://docs.cursor.com/en/context/@-symbols/@-linter-errors

Access and reference linting errors in your codebase

`@Linter Errors` symbol automatically captures and provides context about linting errors and warnings from your currently active file. [Agent](/en/agent/overview) can see lint errors by default.

<Note>
  For linter errors to be visible, you need the appropriate language server
  installed and configured for your programming language. Cursor automatically
  detects and uses installed language servers, but you may need to install
  additional extensions or tools for specific languages.
</Note>

# @Past Chats
Source: https://docs.cursor.com/en/context/@-symbols/@-past-chats

Include summarized chats from history

When working on complex tasks in [Chat](/en/chat), you might need to reference context or decisions from previous conversations. The `@Past Chats` symbol includes summarized versions of previous chats as context.

Particularly useful when:

* You have a long Chat session with important context to reference
* You're starting a new related task and want continuity
* You want to share reasoning or decisions from a previous session

# @Recent Changes
Source: https://docs.cursor.com/en/context/@-symbols/@-recent-changes

Include recently modified code as context

The `@Recent Changes` symbol includes recent code modifications as context in AI conversations.

* Changes are chronologically ordered
* Prioritizes the last 10 changes
* Respects `.cursorignore` settings

# @Web
Source: https://docs.cursor.com/en/context/@-symbols/@-web

Search the web for current information

With `@Web`, Cursor searches the web using [exa.ai](https://exa.ai) to find up-to-date information and add it as context. This includes the ability to parse PDF files from direct links.

<Note>
  Web search is disabled by default. Enable it in Settings → Features → Web
  Search.
</Note>

# Overview
Source: https://docs.cursor.com/en/context/@-symbols/overview

Reference code, files, and documentation using @ symbols

Navigate suggestions using arrow keys. Press `Enter` to select. If the suggestion is a category like `Files`, the suggestions filter to show the most relevant items within that category.

Here's the list of all @ symbols available:

* [@Files](/en/context/@-symbols/@-files) - Reference specific files in your project
* [@Folders](/en/context/@-symbols/@-folders) - Reference entire folders for broader context
* [@Code](/en/context/@-symbols/@-code) - Reference specific code snippets or symbols from your codebase
* [@Docs](/en/context/@-symbols/@-docs) - Access documentation and guides
* [@Git](/en/context/@-symbols/@-git) - Access git history and changes
* [@Past Chats](/en/context/@-symbols/@-past-chats) - Work with summarized composer sessions
* [@Cursor Rules](/en/context/@-symbols/@-cursor-rules) - Work with cursor rules
* [@Web](/en/context/@-symbols/@-web) - Reference external web resources and documentation
* [@Link (paste)](/en/context/@-symbols/@-link) - Create links to specific code or documentation
* [@Recent Changes](/en/context/@-symbols/@-recent-changes) - Create links to specific code or documentation
* [@Lint Errors](/en/context/@-symbols/@-lint-errors) - Reference lint errors ([Chat](/en/chat/overview) only)
* [@Definitions](/en/context/@-symbols/@-definitions) - Look up symbol definitions ([Inline Edit](/en/inline-edit/overview) only)
* [# Files](/en/context/@-symbols/pill-files) - Add files to the context without referencing
* [/ Commands](/en/context/@-symbols/slash-commands) - Add open and active files to the context

# #Files
Source: https://docs.cursor.com/en/context/@-symbols/pill-files

Select specific files using the # prefix

Use `#` followed by a filename to focus on specific files. Combine with `@` symbols for precise context control.

# /command
Source: https://docs.cursor.com/en/context/@-symbols/slash-commands

Quick commands to add files and control context

The `/` command provides quick access to your open editor tabs, letting you add multiple files as context.

## Commands

* **`/Reset Context`**: Resets context to default state
* **`/Generate Cursor Rules`**: Generates rules for Cursor to follow
* **`/Disable Iterate on Lints`**: Will not try to fix linter errors and warnings
* **`/Add Open Files to Context`**: References all editor tabs currently open
* **`/Add Active Files to Context`**: References all editor tabs currently in view (useful with split layouts)

# Codebase Indexing
Source: https://docs.cursor.com/en/context/codebase-indexing

How Cursor learns your codebase for better understanding

Cursor indexes your codebase by computing embeddings for each file. This improves AI-generated answers about your code. When you open a project, Cursor starts indexing automatically. New files are indexed incrementally.
Check indexing status at: `Cursor Settings` > `Indexing & Docs`

## Configuration

Cursor indexes all files except those in [ignore files](/en/context/ignore-files) (e.g. `.gitignore`, `.cursorignore`).

Click `Show Settings` to:

* Enable automatic indexing for new repositories
* Configure which files to ignore

<Tip>
  [Ignoring large content files](/en/context/ignore-files) improves answer
  accuracy.
</Tip>

### View indexed files

To see indexed file paths: `Cursor Settings` > `Indexing & Docs` > `View included files`

This opens a `.txt` file listing all indexed files.

## Multi-root workspaces

Cursor supports [multi-root workspaces](https://code.visualstudio.com/docs/editor/workspaces#_multiroot-workspaces), letting you work with multiple codebases:

* All codebases get indexed automatically
* Each codebase's context is available to AI
* `.cursor/rules` work in all folders

## PR search

PR search helps you understand your codebase's evolution by making historical changes searchable and accessible through AI.

### How it works

Cursor automatically **indexes all merged PRs** from your repository history. Summaries appear in semantic search results, with smart filtering to prioritize recent changes.

Agent can **fetch PRs, commits, issues, or branches** into context using `@[PR number]`, `@[commit hash]`, or `@[branch name]`. Includes GitHub comments and Bugbot reviews when connected.

**Platform support** includes GitHub, GitHub Enterprise, and Bitbucket. GitLab is not currently supported.

<Note>
  GitHub Enterprise users: The fetch tool falls back to git commands due to
  VSCode auth limitations.
</Note>

### Using PR search

Ask questions like "How are services implemented in other PRs?" and Agent will automatically fetch relevant PRs into context to provide comprehensive answers based on your repository's history.

## FAQ

<AccordionGroup>
  <Accordion title="Where can I see all indexed codebases?">
    No global list exists yet. Check each project individually by opening it in
    Cursor and checking Codebase Indexing settings.
  </Accordion>

<Accordion title="How do I delete all indexed codebases?">
    Delete your Cursor account from Settings to remove all indexed codebases.
    Otherwise, delete individual codebases from each project's Codebase Indexing
    settings.
  </Accordion>

<Accordion title="How long are indexed codebases retained?">
    Indexed codebases are deleted after 6 weeks of inactivity. Reopening the
    project triggers re-indexing.
  </Accordion>

<Accordion title="Is my source code stored on Cursor servers?">
    No. Cursor creates embeddings without storing filenames or source code. Filenames are obfuscated and code chunks are encrypted.

When Agent searches the codebase, Cursor retrieves the embeddings from the server and decrypts the chunks.
  </Accordion>
</AccordionGroup>

# Ignore files
Source: https://docs.cursor.com/en/context/ignore-files

Control file access with .cursorignore and .cursorindexingignore

## Overview

Cursor reads and indexes your project's codebase to power its features. Control which directories and files Cursor can access using a `.cursorignore` file in your root directory.

Cursor blocks access to files listed in `.cursorignore` from:

* Codebase indexing
* Code accessible by [Tab](/en/tab/overview), [Agent](/en/agent/overview), and [Inline Edit](/en/inline-edit/overview)
* Code accessible via [@ symbol references](/en/context/@-symbols/overview)

<Warning>
  Tool calls initiated by Agent, like terminal and MCP servers, cannot block
  access to code governed by `.cursorignore`
</Warning>

## Why ignore files?

**Security**: Restrict access to API keys, credentials, and secrets. While Cursor blocks ignored files, complete protection isn't guaranteed due to LLM unpredictability.

**Performance**: In large codebases or monorepos, exclude irrelevant portions for faster indexing and more accurate file discovery.

## Global ignore files

Set ignore patterns for all projects in user settings to exclude sensitive files without per-project configuration.

Default patterns include:

* Environment files: `**/.env`, `**/.env.*`
* Credentials: `**/credentials.json`, `**/secrets.json`
* Keys: `**/*.key`, `**/*.pem`, `**/id_rsa`

## Configuring `.cursorignore`

Create a `.cursorignore` file in your root directory using `.gitignore` syntax.

### Pattern examples

```sh  theme={null}
config.json      # Specific file
dist/           # Directory
*.log           # File extension
**/logs         # Nested directories
!app/           # Exclude from ignore (negate)
```

### Hierarchical ignore

Enable `Cursor Settings` > `Features` > `Editor` > `Hierarchical Cursor Ignore` to search parent directories for `.cursorignore` files.

**Notes**: Comments start with `#`. Later patterns override earlier ones. Patterns are relative to file location.

## Limit indexing with `.cursorindexingignore`

Use `.cursorindexingignore` to exclude files from indexing only. These files remain accessible to AI features but won't appear in codebase searches.

## Files ignored by default

Cursor automatically ignores files in `.gitignore` and the default ignore list below. Override with `!` prefix in `.cursorignore`.

<Accordion title="Default Ignore List">
  For indexing only, these files are ignored in addition to files in your `.gitignore`, `.cursorignore` and `.cursorindexingignore`:

```sh  theme={null}
  package-lock.json
  pnpm-lock.yaml
  yarn.lock
  composer.lock
  Gemfile.lock
  bun.lockb
  .env*
  .git/
  .svn/
  .hg/
  *.lock
  *.bak
  *.tmp
  *.bin
  *.exe
  *.dll
  *.so
  *.lockb
  *.qwoff
  *.isl
  *.csv
  *.pdf
  *.doc
  *.doc
  *.xls
  *.xlsx
  *.ppt
  *.pptx
  *.odt
  *.ods
  *.odp
  *.odg
  *.odf
  *.sxw
  *.sxc
  *.sxi
  *.sxd
  *.sdc
  *.jpg
  *.jpeg
  *.png
  *.gif
  *.bmp
  *.tif
  *.mp3
  *.wav
  *.wma
  *.ogg
  *.flac
  *.aac
  *.mp4
  *.mov
  *.wmv
  *.flv
  *.avi
  *.zip
  *.tar
  *.gz
  *.7z
  *.rar
  *.tgz
  *.dmg
  *.iso
  *.cue
  *.mdf
  *.mds
  *.vcd
  *.toast
  *.img
  *.apk
  *.msi
  *.cab
  *.tar.gz
  *.tar.xz
  *.tar.bz2
  *.tar.lzma
  *.tar.Z
  *.tar.sz
  *.lzma
  *.ttf
  *.otf
  *.pak
  *.woff
  *.woff2
  *.eot
  *.webp
  *.vsix
  *.rmeta
  *.rlib
  *.parquet
  *.svg
  .egg-info/
  .venv/
  node_modules/
  __pycache__/
  .next/
  .nuxt/
  .cache/
  .sass-cache/
  .gradle/
  .DS_Store/
  .ipynb_checkpoints/
  .pytest_cache/
  .mypy_cache/
  .tox/
  .git/
  .hg/
  .svn/
  .bzr/
  .lock-wscript/
  .Python/
  .jupyter/
  .history/
  .yarn/
  .yarn-cache/
  .eslintcache/
  .parcel-cache/
  .cache-loader/
  .nyc_output/
  .node_repl_history/
  .pnp.js/
  .pnp/
  ```
</Accordion>

### Negation pattern limitations

When using negation patterns (prefixed with `!`), you cannot re-include a file if a parent directory is excluded via \*.

```sh  theme={null}
# Ignore all files in public folder
public/*

# ✅ This works, as the file exists at the top level
!public/index.html

# ❌ This doesn't work - cannot re-include files from nested directories
!public/assets/style.css
```

**Workaround**: Explicitly exclude nested directories:

```sh  theme={null}
public/assets/*
!public/assets/style.css # This file is now accessible
```

Excluded directories are not traversed for performance, so patterns on contained files have no effect.
This matches the .gitignore implementation for negation patterns in nested directories. For more details, see the [official Git documentation on gitignore patterns](https://git-scm.com/docs/gitignore).

## Troubleshooting

Test patterns with `git check-ignore -v [file]`.

# Model Context Protocol (MCP)
Source: https://docs.cursor.com/en/context/mcp

Connect external tools and data sources to Cursor using MCP

## What is MCP?

[Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) enables Cursor to connect to external tools and data sources.

### Why use MCP?

MCP connects Cursor to external systems and data. Instead of explaining your project structure repeatedly, integrate directly with your tools.

Write MCP servers in any language that can print to `stdout` or serve an HTTP endpoint - Python, JavaScript, Go, etc.

### How it works

MCP servers expose capabilities through the protocol, connecting Cursor to external tools or data sources.

Cursor supports three transport methods:

### Protocol support

Cursor supports these MCP protocol capabilities:

## Installing MCP servers

### One-click installation

Install MCP servers from our collection and authenticate with OAuth.

<Columns cols={2}>
  <Card title="Browse MCP Tools" icon="table" horizontal href="/en/tools">
    Browse available MCP servers
  </Card>

<Card title="Add to Cursor Button" icon="plus" horizontal href="/en/deeplinks">
    Create an "Add to Cursor" button
  </Card>
</Columns>

### Using `mcp.json`

Configure custom MCP servers with a JSON file:

<CodeGroup>
  ```json CLI Server - Node.js theme={null}
  {
    "mcpServers": {
      "server-name": {
        "command": "npx",
        "args": ["-y", "mcp-server"],
        "env": {
          "API_KEY": "value"
        }
      }
    }
  }
  ```

```json CLI Server - Python theme={null}
  {
    "mcpServers": {
      "server-name": {
        "command": "python",
        "args": ["mcp-server.py"],
        "env": {
          "API_KEY": "value"
        }
      }
    }
  }
  ```

```json Remote Server theme={null}
  // MCP server using HTTP or SSE - runs on a server
  {
    "mcpServers": {
      "server-name": {
        "url": "http://localhost:3000/mcp",
        "headers": {
          "API_KEY": "value"
        }
      }
    }
  }
  ```
</CodeGroup>

### STDIO server configuration

For STDIO servers (local command-line servers), configure these fields in your `mcp.json`:

<div className="full-width-table">
  | Field       | Required | Description                                                                                             | Examples                                  |
  | :---------- | :------- | :------------------------------------------------------------------------------------------------------ | :---------------------------------------- |
  | **type**    | Yes      | Server connection type                                                                                  | `"stdio"`                                 |
  | **command** | Yes      | Command to start the server executable. Must be available on your system path or contain its full path. | `"npx"`, `"node"`, `"python"`, `"docker"` |
  | **args**    | No       | Array of arguments passed to the command                                                                | `["server.py", "--port", "3000"]`         |
  | **env**     | No       | Environment variables for the server                                                                    | `{"API_KEY": "${input:api-key}"}`         |
  | **envFile** | No       | Path to an environment file to load more variables                                                      | `".env"`, `"${workspaceFolder}/.env"`     |
</div>

### Using the Extension API

For programmatic MCP server registration, Cursor provides an extension API that allows dynamic configuration without modifying `mcp.json` files. This is particularly useful for enterprise environments and automated setup workflows.

<Card title="MCP Extension API Reference" icon="code" href="/en/context/mcp-extension-api">
  Learn how to register MCP servers programmatically using `vscode.cursor.mcp.registerServer()`
</Card>

### Configuration locations

<CardGroup cols={2}>
  <Card title="Project Configuration" icon="folder-tree">
    Create `.cursor/mcp.json` in your project for project-specific tools.
  </Card>

<Card title="Global Configuration" icon="globe">
    Create `~/.cursor/mcp.json` in your home directory for tools available everywhere.
  </Card>
</CardGroup>

### Config interpolation

Use variables in `mcp.json` values. Cursor resolves variables in these fields: `command`, `args`, `env`, `url`, and `headers`.

Supported syntax:

* `${env:NAME}` environment variables
* `${userHome}` path to your home folder
* `${workspaceFolder}` project root (the folder that contains `.cursor/mcp.json`)
* `${workspaceFolderBasename}` name of the project root
* `${pathSeparator}` and `${/}` OS path separator

Examples

```json  theme={null}
{
  "mcpServers": {
    "local-server": {
      "command": "python",
      "args": ["${workspaceFolder}/tools/mcp_server.py"],
      "env": {
        "API_KEY": "${env:API_KEY}"
      }
    }
  }
}
```

```json  theme={null}
{
  "mcpServers": {
    "remote-server": {
      "url": "https://api.example.com/mcp",
      "headers": {
        "Authorization": "Bearer ${env:MY_SERVICE_TOKEN}"
      }
    }
  }
}
```

### Authentication

MCP servers use environment variables for authentication. Pass API keys and tokens through the config.

Cursor supports OAuth for servers that require it.

## Using MCP in chat

The Composer Agent automatically uses MCP tools listed under `Available Tools` when relevant. Ask for a specific tool by name or describe what you need. Enable or disable tools from settings.

### Toggling tools

Enable or disable MCP tools directly from the chat interface. Click a tool name in the tools list to toggle it. Disabled tools won't be loaded into context or available to Agent.

### Tool approval

Agent asks for approval before using MCP tools by default. Click the arrow next to the tool name to see arguments.

#### Auto-run

Enable auto-run for Agent to use MCP tools without asking. Works like terminal commands. Read more about Auto-run settings [here](/en/agent/tools#auto-run).

### Tool response

Cursor shows the response in chat with expandable views of arguments and responses:

### Images as context

MCP servers can return images - screenshots, diagrams, etc. Return them as base64 encoded strings:

```js  theme={null}
const RED_CIRCLE_BASE64 = "/9j/4AAQSkZJRgABAgEASABIAAD/2w...";
// ^ full base64 clipped for readability

server.tool("generate_image", async (params) => {
  return {
    content: [
      {
        type: "image",
        data: RED_CIRCLE_BASE64,
        mimeType: "image/jpeg",
      },
    ],
  };
});
```

See this [example server](https://github.com/msfeldstein/mcp-test-servers/blob/main/src/image-server.js) for implementation details. Cursor attaches returned images to the chat. If the model supports images, it analyzes them.

## Security considerations

When installing MCP servers, consider these security practices:

* **Verify the source**: Only install MCP servers from trusted developers and repositories
* **Review permissions**: Check what data and APIs the server will access
* **Limit API keys**: Use restricted API keys with minimal required permissions
* **Audit code**: For critical integrations, review the server's source code

Remember that MCP servers can access external services and execute code on your behalf. Always understand what a server does before installation.

## Real-world examples

For practical examples of MCP in action, see our [Web Development guide](/en/guides/tutorials/web-development) which demonstrates integrating Linear, Figma, and browser tools into your development workflow.

## FAQ

<AccordionGroup>
  <Accordion title="What's the point of MCP servers?">
    MCP servers connect Cursor to external tools like Google Drive, Notion, and
    other services to bring docs and requirements into your coding workflow.
  </Accordion>

{" "}

<Accordion title="How do I debug MCP server issues?">
    View MCP logs by: 1. Open the Output panel in Cursor (<Kbd>Cmd+Shift+U</Kbd>)
    2\. Select "MCP Logs" from the dropdown 3. Check for connection errors,
    authentication issues, or server crashes The logs show server initialization,
    tool calls, and error messages.
  </Accordion>

{" "}

<Accordion title="Can I temporarily disable an MCP server?">
    Yes! Toggle servers on/off without removing them: 1. Open Settings (
    <Kbd>Cmd+Shift+J</Kbd>) 2. Go to Features → Model Context Protocol 3. Click
    the toggle next to any server to enable/disable Disabled servers won't load or
    appear in chat. This is useful for troubleshooting or reducing tool clutter.
  </Accordion>

{" "}

<Accordion title="What happens if an MCP server crashes or times out?">
    If an MCP server fails: - Cursor shows an error message in chat - The tool
    call is marked as failed - You can retry the operation or check logs for
    details - Other MCP servers continue working normally Cursor isolates server
    failures to prevent one server from affecting others.
  </Accordion>

{" "}

<Accordion title="How do I update an MCP server?">
    For npm-based servers: 1. Remove the server from settings 2. Clear npm cache:
    `npm cache clean --force` 3. Re-add the server to get the latest version For
    custom servers, update your local files and restart Cursor.
  </Accordion>

<Accordion title="Can I use MCP servers with sensitive data?">
    Yes, but follow security best practices: - Use environment variables for
    secrets, never hardcode them - Run sensitive servers locally with `stdio`
    transport - Limit API key permissions to minimum required - Review server
    code before connecting to sensitive systems - Consider running servers in
    isolated environments
  </Accordion>
</AccordionGroup>

# Memories
Source: https://docs.cursor.com/en/context/memories

Memories are automatically generated rules based on your conversations in Chat. These memories are scoped to your project and maintain context across sessions.

## How memories are created

1. **Sidecar observation**: Cursor uses a sidecar approach where another model observes your conversations and automatically extracts relevant memories. This happens passively in the background as you work. Background-generated memories require user approval before being saved, ensuring trust and control over what gets remembered.

2. **Tool calls**: Agent can directly create memories using tool calls when you explicitly ask it to remember something or when it notices important information that should be preserved for future sessions.

## Manage memories

You can manage memories from Cursor Settings → Rules.

# Rules
Source: https://docs.cursor.com/en/context/rules

Control how the Agent model behaves with reusable, scoped instructions.

Rules provide system-level instructions to Agent and Inline Edit. Think of them as persistent context, preferences, or workflows for your projects.

Cursor supports four types of rules:

<CardGroup cols={2}>
  <Card title="Project Rules" icon="folder-tree">
    Stored in `.cursor/rules`, version-controlled and scoped to your codebase.
  </Card>

<Card title="User Rules" icon="user">
    Global to your Cursor environment. Defined in settings and always applied.
  </Card>

<Card title="AGENTS.md" icon="robot">
    Agent instructions in markdown format. Simple alternative to `.cursor/rules`.
  </Card>

<Card title=".cursorrules (Legacy)" icon="clock-rotate-left">
    Still supported, but deprecated. Use Project Rules instead.
  </Card>
</CardGroup>

## How rules work

Large language models don't retain memory between completions. Rules provide persistent, reusable context at the prompt level.

When applied, rule contents are included at the start of the model context. This gives the AI consistent guidance for generating code, interpreting edits, or helping with workflows.

<Info>
  Rules apply to [Chat](/en/chat/overview) and [Inline
  Edit](/en/inline-edit/overview). Active rules show in the Agent sidebar.
</Info>

## Project rules

Project rules live in `.cursor/rules`. Each rule is a file and version-controlled. They can be scoped using path patterns, invoked manually, or included based on relevance. Subdirectories can include their own `.cursor/rules` directory scoped to that folder.

Use project rules to:

* Encode domain-specific knowledge about your codebase
* Automate project-specific workflows or templates
* Standardize style or architecture decisions

### Rule anatomy

Each rule file is written in **MDC** (`.mdc`), a format supporting metadata and content. Control how rules are applied from the type dropdown which changes properties `description`, `globs`, `alwaysApply`.

| <span class="no-wrap">Rule Type</span>         | Description                                                                      |
| :--------------------------------------------- | :------------------------------------------------------------------------------- |
| <span class="no-wrap">`Always`</span>          | Always included in model context                                                 |
| <span class="no-wrap">`Auto Attached`</span>   | Included when files matching a glob pattern are referenced                       |
| <span class="no-wrap">`Agent Requested`</span> | Available to AI, which decides whether to include it. Must provide a description |
| <span class="no-wrap">`Manual`</span>          | Only included when explicitly mentioned using `@ruleName`                        |

```
---
description: RPC Service boilerplate
globs:
alwaysApply: false
---

- Use our internal RPC pattern when defining services
- Always use snake_case for service names.

@service-template.ts
```

### Nested rules

Organize rules by placing them in `.cursor/rules` directories throughout your project. Nested rules automatically attach when files in their directory are referenced.

```
project/
  .cursor/rules/        # Project-wide rules
  backend/
    server/
      .cursor/rules/    # Backend-specific rules
  frontend/
    .cursor/rules/      # Frontend-specific rules
```

### Creating a rule

Create rules using the `New Cursor Rule` command or going to `Cursor Settings > Rules`. This creates a new rule file in `.cursor/rules`. From settings you can see all rules and their status.

### Generating rules

Generate rules directly in conversations using the `/Generate Cursor Rules` command. Useful when you've made decisions about agent behavior and want to reuse them.

<Frame>
  <video src="https://www.cursor.com/changelog/049/generate-rules.mp4" controls>
    Your browser does not support the video tag.
  </video>
</Frame>

## Best practices

Good rules are focused, actionable, and scoped.

* Keep rules under 500 lines
* Split large rules into multiple, composable rules
* Provide concrete examples or referenced files
* Avoid vague guidance. Write rules like clear internal docs
* Reuse rules when repeating prompts in chat

## Examples

<AccordionGroup>
  <Accordion title="Standards for frontend components and API validation">
    This rule provides standards for frontend components:

When working in components directory:

* Always use Tailwind for styling
    * Use Framer Motion for animations
    * Follow component naming conventions

This rule enforces validation for API endpoints:

In API directory:

* Use zod for all validation
    * Define return types with zod schemas
    * Export types generated from schemas
  </Accordion>

<Accordion title="Templates for Express services and React components">
    This rule provides a template for Express services:

Use this template when creating Express service:

* Follow RESTful principles
    * Include error handling middleware
    * Set up proper logging

@express-service-template.ts

This rule defines React component structure:

React components should follow this layout:

* Props interface at top
    * Component as named export
    * Styles at bottom

@component-template.tsx
  </Accordion>

<Accordion title="Automating development workflows and documentation generation">
    This rule automates app analysis:

When asked to analyze the app:

1. Run dev server with `npm run dev`
    2. Fetch logs from console
    3. Suggest performance improvements

This rule helps generate documentation:

Help draft documentation by:

* Extracting code comments
    * Analyzing README.md
    * Generating markdown documentation
  </Accordion>

<Accordion title="Adding a new setting in Cursor">
    First create a property to toggle in `@reactiveStorageTypes.ts`.

Add default value in `INIT_APPLICATION_USER_PERSISTENT_STORAGE` in `@reactiveStorageService.tsx`.

For beta features, add toggle in `@settingsBetaTab.tsx`, otherwise add in `@settingsGeneralTab.tsx`. Toggles can be added as `<SettingsSubSection>` for general checkboxes. Look at the rest of the file for examples.

```
    <SettingsSubSection
    				label="Your feature name"
    				description="Your feature description"
    				value={
    					vsContext.reactiveStorageService.applicationUserPersistentStorage
    						.myNewProperty ?? false
    				}
    				onChange={(newVal) => {
    					vsContext.reactiveStorageService.setApplicationUserPersistentStorage(
    						'myNewProperty',
    						newVal
    					);
    				}}
    			/>
    ```

To use in the app, import reactiveStorageService and use the property:

```
    const flagIsEnabled = vsContext.reactiveStorageService.applicationUserPersistentStorage.myNewProperty
    ```
  </Accordion>
</AccordionGroup>

Many examples available from providers and frameworks. Community-contributed rules are found across crowdsourced collections and repositories online.

## AGENTS.md

`AGENTS.md` is a simple markdown file for defining agent instructions. Place it in your project root as an alternative to `.cursor/rules` for straightforward use cases.

Unlike Project Rules, `AGENTS.md` is a plain markdown file without metadata or complex configurations. It's perfect for projects that need simple, readable instructions without the overhead of structured rules.

```markdown  theme={null}
# Project Instructions

## Code Style
- Use TypeScript for all new files
- Prefer functional components in React
- Use snake_case for database columns

## Architecture
- Follow the repository pattern
- Keep business logic in service layers
```

## User Rules

User rules are global preferences defined in **Cursor Settings → Rules** that apply across all projects. They're plain text and perfect for setting preferred communication style or coding conventions:

```
Please reply in a concise style. Avoid unnecessary repetition or filler language.
```

## `.cursorrules` (Legacy)

The `.cursorrules` file in your project root is still supported but will be deprecated. We recommend migrating to Project Rules for more control, flexibility, and visibility.

## FAQ

<AccordionGroup>
  <Accordion title="Why isn't my rule being applied?">
    Check the rule type. For `Agent Requested`, ensure a description is defined. For `Auto Attached`, ensure the file pattern matches referenced files.
  </Accordion>

{" "}

<Accordion title="Can rules reference other rules or files?">
    Yes. Use `@filename.ts` to include files in your rule's context.
  </Accordion>

{" "}

<Accordion title="Can I create a rule from chat?">
    Yes, generate project rules from chat using the `/Generate Cursor Rules`
    command. If Memories are enabled, memories are generated automatically.
  </Accordion>

<Accordion title="Do rules impact Cursor Tab or other AI features?">
    No. Rules only apply to Agent and Inline Edit
  </Accordion>
</AccordionGroup>

# Concepts
Source: https://docs.cursor.com/en/get-started/concepts

Learn the key features that make Cursor powerful

<p className="text-sm">
        Code completion that predicts multi-line edits. Press Tab to accept
        suggestions based on your current code and recent changes.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/agent/overview" className="hover:text-primary transition-colors">
          Agent
        </a>
      </h3>

<p className="text-sm">
        An AI that can read and modify code across multiple files. Describe
        changes in natural language and Agent executes them.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/background-agent" className="hover:text-primary transition-colors">
          Background Agent
        </a>
      </h3>

<p className="text-sm">
        Runs tasks asynchronously while you continue working. Can be accessed
        through the editor or external integrations like Slack.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/inline-edit/overview" className="hover:text-primary transition-colors">
          Inline Edit
        </a>
      </h3>

<p className="text-sm">
        Edit selected code with natural language. Press <Kbd>Cmd+K</Kbd> to
        describe changes and see them applied in place.
      </p>
    </div>

<p className="text-sm">
        Interface for AI conversations. Supports multiple tabs, conversation
        history, checkpoints, and export functionality.
      </p>
    </div>

<p className="text-sm">
        Custom instructions that define AI behavior. Set coding standards,
        framework preferences, and project-specific conventions.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/context/memories" className="hover:text-primary transition-colors">
          Memories
        </a>
      </h3>

<p className="text-sm">
        Persistent storage of project context and decisions from past
        conversations. Automatically referenced in future interactions.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/context/codebase-indexing" className="hover:text-primary transition-colors">
          Codebase Indexing
        </a>
      </h3>

<p className="text-sm">
        Semantic analysis of your codebase. Enables code search, reference
        finding, and context-aware suggestions.
      </p>
    </div>

<p className="text-sm">
        Model Context Protocol for integrating external tools. Connects to
        databases, APIs, and documentation sources.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/guides/working-with-context" className="hover:text-primary transition-colors">
          Context
        </a>
      </h3>

<p className="text-sm">
        Information provided to AI models during code generation. Includes files,
        symbols, and conversation history.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/en/models" className="hover:text-primary transition-colors">
          Models
        </a>
      </h3>

<p className="text-sm">
        Different AI models available for code generation. Each model has
        different speed and capability characteristics.
      </p>
    </div>

# Installation
Source: https://docs.cursor.com/en/get-started/installation

Get Cursor installed on your computer in just a few minutes

## Download Cursor

Getting started is simple:

1. Go to [cursor.com](https://cursor.com) and click "Download"
2. Run the installer once it downloads
3. Open Cursor when installation finishes

<Info>
  Need a specific version? Find all platforms and installation methods at
  [cursor.com/downloads](https://cursor.com/downloads)
</Info>

## First-time setup

When you open Cursor for the first time, we'll walk you through a quick setup:

* Choose keyboard shortcuts that feel familiar
* Pick a theme you like
* Set up your terminal preferences

Your browser does not support the video tag.
  </video>
</Frame>

<Tip>
  You can revisit the setup wizard anytime by pressing <Kbd>Cmd Shift P</Kbd>{" "}
  and searching for `Cursor: Start Onboarding`.
</Tip>

Read more about [Keyboard Shortcuts](/en/kbd), [Themes](/en/settings/themes) and [Shell Commands](/en/settings/shell)

<CardGroup cols={3}>
  <Card title="Keyboard shortcuts" href="/en/configuration/kbd" arrow>
    See keyboard shortcuts
  </Card>

<Card title="Themes" href="/en/configuration/themes" arrow>
    Selecting a theme in Cursor
  </Card>

<Card title="Shell Commands" href="/en/configuration/shell" arrow>
    Install shell commands
  </Card>
</CardGroup>

## Moving from another editor?

If you're already using another code editor, we make it easy to switch:

<CardGroup cols={2}>
  <Card title="VS Code" href="/en/guides/migration/vscode" arrow>
    Import VS Code settings directly
  </Card>

<Card title="Jetbrains" href="/en/guides/migration/jetbrains" arrow>
    Migration guides for JetBrains, Eclipse, Neovim, and Sublime
  </Card>
</CardGroup>

More migration guides coming soon.

## Language support

Cursor works with all major programming languages. Here are some popular ones with enhanced AI support:

<CardGroup cols={4}>
  <Card
    title="TypeScript"
    href="/en/guides/languages/javascript"
    icon={<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 512 512" fill="none">
  <rect width={512} height={512} fill="#3178c6" rx={50} />
  <rect width={512} height={512} fill="#3178c6" rx={50} />
  <path
    fill="#fff"
    fillRule="evenodd"
    d="M316.939 407.424v50.061c8.138 4.172 17.763 7.3 28.875 9.386S368.637 470 380.949 470c11.999 0 23.397-1.147 34.196-3.442 10.799-2.294 20.268-6.075 28.406-11.342 8.138-5.266 14.581-12.15 19.328-20.65S470 415.559 470 403.044c0-9.074-1.356-17.026-4.069-23.857s-6.625-12.906-11.738-18.225c-5.112-5.319-11.242-10.091-18.389-14.315s-15.207-8.213-24.18-11.967c-6.573-2.712-12.468-5.345-17.685-7.9-5.217-2.556-9.651-5.163-13.303-7.822-3.652-2.66-6.469-5.476-8.451-8.448-1.982-2.973-2.974-6.336-2.974-10.091 0-3.441.887-6.544 2.661-9.308s4.278-5.136 7.512-7.118c3.235-1.981 7.199-3.52 11.894-4.615 4.696-1.095 9.912-1.642 15.651-1.642 4.173 0 8.581.313 13.224.938 4.643.626 9.312 1.591 14.008 2.894a97.514 97.514 0 0 1 13.694 4.928c4.434 1.982 8.529 4.276 12.285 6.884v-46.776c-7.616-2.92-15.937-5.084-24.962-6.492S415.797 238 404.112 238c-11.895 0-23.163 1.278-33.805 3.833s-20.006 6.544-28.093 11.967c-8.086 5.424-14.476 12.333-19.171 20.729-4.695 8.395-7.043 18.433-7.043 30.114 0 14.914 4.304 27.638 12.912 38.172 8.607 10.533 21.675 19.45 39.204 26.751 6.886 2.816 13.303 5.579 19.25 8.291s11.086 5.528 15.415 8.448c4.33 2.92 7.747 6.101 10.252 9.543 2.504 3.441 3.756 7.352 3.756 11.733 0 3.233-.783 6.231-2.348 8.995s-3.939 5.162-7.121 7.196-7.147 3.624-11.894 4.771c-4.748 1.148-10.303 1.721-16.668 1.721-10.851 0-21.597-1.903-32.24-5.71-10.642-3.806-20.502-9.516-29.579-17.13zM232.78 284.082H297V243H118v41.082h63.906V467h50.874z"
    clipRule="evenodd"
  />
</svg>}
    arrow
  />

<Card
    title="Java"
    href="/en/guides/languages/java"
    icon={ <svg
  xmlns="http://www.w3.org/2000/svg"
  fill="none"
  aria-label="Java"
  viewBox="0 0 512 512"
  width="32"

</svg>}
    arrow
  />

<Card
    title="Python"
    href="/en/guides/languages/python"
    icon={
  <svg
  xmlns="http://www.w3.org/2000/svg"
  width="24"
  height="24"
  fill="none"
  viewBox="0 0 32 32"
>
  <path
    fill="url(#a)"
    fillRule="evenodd"
    d="M13.016 2C10.82 2 9.038 3.725 9.038 5.852v2.667h6.886v.74H5.978C3.781 9.26 2 10.984 2 13.111v5.778c0 2.127 1.781 3.852 3.978 3.852h2.295v-3.26c0-2.127 1.781-3.851 3.978-3.851h7.345c1.859 0 3.366-1.46 3.366-3.26V5.852C22.962 3.725 21.18 2 18.984 2h-5.968Zm-.918 4.74c.76 0 1.377-.596 1.377-1.333 0-.736-.616-1.333-1.377-1.333-.76 0-1.377.597-1.377 1.333 0 .737.617 1.334 1.377 1.334Z"
    clipRule="evenodd"
  />
  <path
    fill="url(#b)"
    fillRule="evenodd"
    d="M18.983 30c2.197 0 3.979-1.724 3.979-3.852v-2.666h-6.886v-.741h9.946c2.197 0 3.978-1.725 3.978-3.852V13.11c0-2.127-1.781-3.852-3.978-3.852h-2.295v3.26c0 2.127-1.782 3.851-3.979 3.851h-7.344c-1.859 0-3.366 1.46-3.366 3.26v6.518c0 2.128 1.781 3.852 3.978 3.852h5.967Zm.918-4.74c-.76 0-1.377.596-1.377 1.333 0 .736.617 1.333 1.377 1.333.761 0 1.378-.597 1.378-1.333 0-.737-.617-1.334-1.378-1.334Z"
    clipRule="evenodd"
  />
  <defs>
    <linearGradient
      id="a"
      x1={12.481}
      x2={12.481}
      y1={2}
      y2={22.741}
      gradientUnits="userSpaceOnUse"
    >
      <stop stopColor="#327EBD" />
      <stop offset={1} stopColor="#1565A7" />
    </linearGradient>
    <linearGradient
      id="b"
      x1={19.519}
      x2={19.519}
      y1={9.259}
      y2={30}
      gradientUnits="userSpaceOnUse"
    >
      <stop stopColor="#FFDA4B" />
      <stop offset={1} stopColor="#F9C600" />
    </linearGradient>
  </defs>
</svg>
}
    arrow
  />

<Card
    title="Swift"
    href="/en/guides/languages/swift"
    icon={
    <svg
  xmlns="http://www.w3.org/2000/svg"
  xmlSpace="preserve"
  width="24"
  height="24"
  viewBox="0 0 59.391 59.391"
>
  <path
    fill="#F05138"
    d="M59.387 16.45a82.463 82.463 0 0 0-.027-1.792c-.035-1.301-.112-2.614-.343-3.9-.234-1.307-.618-2.523-1.222-3.71a12.464 12.464 0 0 0-5.453-5.452C51.156.992 49.941.609 48.635.374c-1.288-.232-2.6-.308-3.902-.343a85.714 85.714 0 0 0-1.792-.027C42.23 0 41.52 0 40.813 0H18.578c-.71 0-1.419 0-2.128.004-.597.004-1.195.01-1.792.027-.325.009-.651.02-.978.036-.978.047-1.959.133-2.924.307-.98.176-1.908.436-2.811.81A12.503 12.503 0 0 0 3.89 3.89a12.46 12.46 0 0 0-2.294 3.158C.992 8.235.61 9.45.374 10.758c-.231 1.286-.308 2.599-.343 3.9a85.767 85.767 0 0 0-.027 1.792C-.001 17.16 0 17.869 0 18.578v22.235c0 .71 0 1.418.004 2.128.004.597.01 1.194.027 1.791.035 1.302.112 2.615.343 3.901.235 1.307.618 2.523 1.222 3.71a12.457 12.457 0 0 0 5.453 5.453c1.186.603 2.401.986 3.707 1.22 1.287.232 2.6.31 3.902.344.597.016 1.195.023 1.793.027.709.005 1.417.004 2.127.004h22.235c.709 0 1.418 0 2.128-.004.597-.004 1.194-.011 1.792-.027 1.302-.035 2.614-.112 3.902-.343 1.306-.235 2.521-.618 3.707-1.222a12.461 12.461 0 0 0 5.453-5.452c.604-1.187.987-2.403 1.222-3.71.231-1.286.308-2.6.343-3.9.016-.598.023-1.194.027-1.792.004-.71.004-1.419.004-2.129V18.578c0-.71 0-1.419-.004-2.128z"
  />
  <path
    fill="#FFF"
    d="m47.06 36.66-.004-.004c.066-.224.134-.446.191-.675 2.465-9.821-3.55-21.432-13.731-27.546 4.461 6.048 6.434 13.374 4.681 19.78-.156.571-.344 1.12-.552 1.653-.225-.148-.51-.316-.89-.527 0 0-10.127-6.252-21.103-17.312-.288-.29 5.852 8.777 12.822 16.14-3.284-1.843-12.434-8.5-18.227-13.802.712 1.187 1.558 2.33 2.489 3.43C17.573 23.932 23.882 31.5 31.44 37.314c-5.31 3.25-12.814 3.502-20.285.003a30.646 30.646 0 0 1-5.193-3.098c3.162 5.058 8.033 9.423 13.96 11.97 7.07 3.039 14.1 2.833 19.336.05l-.004.007c.024-.016.055-.032.08-.047.214-.116.428-.234.636-.358 2.516-1.306 7.485-2.63 10.152 2.559.654 1.27 2.041-5.46-3.061-11.74z"
  />
</svg>
}
    arrow
  />
</CardGroup>

You can add support for more languages through extensions, just like VS Code.

## Creating your account

While Cursor works without an account, signing up unlocks all AI features:

1. You'll be prompted to sign up during setup, or you can do it later in Settings (<Kbd>Cmd Shift J</Kbd>)
2. Once signed up, manage your account at [cursor.com/dashboard](https://cursor.com/dashboard)

## Understanding codebase indexing

When you open a project, Cursor starts learning about your code. This is called "indexing" and it's what makes the AI suggestions accurate.

* Indexing starts automatically when you open a project
* It takes about 1-15 minutes depending on your project size
* The more Cursor learns about your code, the smarter its suggestions become
* Teams can share indexes with each other to save time
* You can check indexing progress in Settings (<Kbd>Cmd Shift J</Kbd>) → Indexing & Docs

Want to learn more? See [how indexing works](/en/context/codebase-indexing)

## Next steps

Now that Cursor is installed, you're ready to experience AI-powered coding:

* Follow our [quickstart guide](/en/get-started/quickstart) to learn the basics in 5 minutes
* Read about [key concepts](/en/get-started/concepts) to understand how Cursor works
* [Browse guides](/en/guides) to explore what you can build with Cursor
* If you run into issues, [get help](/en/troubleshooting/common-issues) from our troubleshooting guide
* [Join our community](https://cursor.com/community) to connect with other Cursor users

# Quickstart
Source: https://docs.cursor.com/en/get-started/quickstart

Get started with Cursor in 5 minutes

This quickstart will walk you through a project using Cursor's core features. By the end, you'll be familiar with Tab, Inline Edit, and Agent.

## Open a project in Cursor

Use an existing project or clone our example:

<Tabs>
  <Tab title="Clone example project">
    1. Ensure git is installed
    2. Clone the example project:

```bash  theme={null}
    git clone git@github.com:voxelize/voxelize.git && \
    cd voxelize && \
    cursor .
    ```
  </Tab>

<Tab title="Use existing project">
    1. Open Cursor
    2. Open a project folder with <Kbd>Cmd O</Kbd> or `cursor <path-to-project>`
  </Tab>
</Tabs>

We'll be showcasing using the example project, but you can use any project you have locally.

## Autocomplete with [Tab](/en/kbd#tab)

Tab is the autocomplete model we've trained in-house. It's a great way to ease into AI assisted coding if you're not used to it. With Tab, you can:

* Autocomplete **multiple lines and blocks** of code
* Jump **in** and **across** files to the next autocomplete suggestion

1. Start typing the beginning of a function:
   ```javascript  theme={null}
   function calculate
   ```
2. Tab suggestions appear automatically
3. Press Tab to accept the suggestion
4. Cursor suggests parameters and function bodies

## [Inline Edit](/en/inline-edit) a selection

1. Select the function you just created
2. Press <Kbd>Cmd K</Kbd>
3. Type "make this function calculate fibonacci numbers"
4. Press <Kbd>Return</Kbd> to apply the changes
5. Cursor adds imports and documentation

## Chat with [Agent](/en/agent)

1. Open the Chat panel (<Kbd>Cmd I</Kbd>)
2. Ask: "Add tests for this function and run them"
3. Agent will create a test file, write test cases, and run them for you

## Bonus

Advanced features:

<AccordionGroup>
  <Accordion title="Handoff work to Background Agent">
    1. Open the Background Agent control panel (<Kbd>Cmd E</Kbd>)
    2. Ask: "Find and fix a bug in this project"
    3. [Background Agent](/en/background-agent) will:
       * Create a remote Virtual Machine (VM)
       * Explore your project
       * Detect bugs
       * Propose fixes

Review and apply changes.
  </Accordion>

{" "}

<Accordion title="Write a rule">
    1. Open the command palette (<Kbd>Cmd Shift P</Kbd>) 2. Search: "New Cursor
       Rule" 3. Name it (e.g., `style-guide`) 4. Select Rule Type "Always" 5. Define
       your style: `Prefer using camelCase for variable names`
  </Accordion>

<Accordion title="Set up an MCP server">
    1. Visit our [MCP directory](https://docs.cursor.com/tools)
    2. Choose a tool
    3. Click "Install"

Servers can also be installed manually:

1. Open Cursor Settings (<Kbd>Cmd Shift J</Kbd>)
    2. Go to "Tools & Integrations"
    3. Click "New MCP Server"
  </Accordion>
</AccordionGroup>

## Next steps

Explore these guides to learn more:

<CardGroup cols={2}>
  <Card title="Working with Context" href="/en/guides/working-with-context">
    Provide effective context for better results
  </Card>

<Card title="Selecting Models" href="/en/guides/selecting-models">
    Choose the right model for your task
  </Card>
</CardGroup>

Learn all [Cursor concepts](/en/get-started/concepts) and start building!

# Data Science
Source: https://docs.cursor.com/en/guides/advanced/datascience

Learn how to set up Cursor for data science workflows including Python, R, and SQL with notebooks, remote environments, and AI-powered analysis

Cursor provides integrated tooling for data science development through reproducible environments, notebook
support, and AI-powered code assistance. This guide covers essential setup patterns for Python, R, and SQL
workflows.

## Notebook development

<Note>
  For full notebook support, download the Jupyter (id: ms-toolsai.jupyter) extension, published by ms-toolsai.
</Note>

Cursor supports both `.ipynb` and `.py` files with integrated cell execution. Tab, Inline Edit, and Agents
work within notebooks, just as they do in other code files.

Key capabilities:

* **Inline cell execution** runs code directly within the editor interface
* **Tab, Inline Edit, and Agent** all understand data science libraries including pandas, NumPy, scikit-learn, and SQL magic commands

## Database integration

Databases can be integrated with Cursor through two main mechanisms: MCP servers and Extensions.

* **MCP Servers** let your Agents connect with your databases
* **Extensions** integrate your broader IDE with your databases

### Via MCP

MCP servers allow your agent to make queries directly against your database. This allows your agent to choose to query your database, write the appropriate query, run the command and analyze outputs, all as part of an ongoing task.

For example, you can connect a Postgres database to your Cursor instance by adding the following [MCP config](https://github.com/modelcontextprotocol/servers-archived/tree/main/src/postgres) to Cursor:

```json  theme={null}
{
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-postgres",
        "postgresql://localhost/mydb"
      ]
    }
  }
}
```

For more on MCP, see our [MCP documentation](/en/tools/mcp).

Your browser does not support the video tag.
  </video>
</Frame>

### Via Extensions

Install database-specific extensions (PostgreSQL, BigQuery, SQLite, Snowflake) to execute queries directly from the editor. This eliminates context switching between tools and enables AI assistance for query optimization.

```sql  theme={null}
-- Cursor provides suggestions for indexes, window functions, and query optimization
SELECT
    user_id,
    event_type,
    COUNT(*) as event_count,
    RANK() OVER (PARTITION BY user_id ORDER BY COUNT(*) DESC) as frequency_rank
FROM events
WHERE created_at >= NOW() - INTERVAL '7 days'
GROUP BY user_id, event_type;
```

Use Agents to analyze slow queries, suggest performance improvements, or generate visualization code for query results. Cursor understands SQL context and can recommend appropriate chart types based on your data structure.

## Data visualization

Cursor's AI assistance extends to data visualization libraries including Matplotlib, Plotly, and Seaborn. The agent can generate code for data visualization, helping you quickly and easily explore data, while creating a replicable and shareable artifact.

```python  theme={null}
import plotly.express as px
import pandas as pd

# AI suggests relevant plot types based on data columns
df = pd.read_csv('sales_data.csv')
fig = px.scatter(df, x='advertising_spend', y='revenue',
                 color='region', size='customer_count',
                 title='Revenue vs Advertising Spend by Region')
fig.show()
```

Your browser does not support the video tag.
  </video>
</Frame>

## Frequently asked questions

**Can I use existing Jupyter notebooks?**
Yes, Cursor opens `.ipynb` files with full cell execution and AI completion support.

**How do I handle large datasets that don't fit in memory?**
Use distributed computing libraries like Dask or connect to Spark clusters through Remote-SSH connections to larger machines.

**Does Cursor support R and SQL files?**
Yes, Cursor provides AI assistance and syntax highlighting for R scripts (`.R`) and SQL files (`.sql`).

**What's the recommended way to share development environments?**
Commit the `.devcontainer` folder to version control. Team members can rebuild the environment automatically when opening the project.

**How do I debug data processing pipelines?**
Use Cursor's integrated debugger with breakpoints in Python scripts, or leverage Agent to analyze and explain complex data transformations step by step.

## Environment reproducibility

### Development containers

Development containers help you ensure consistent runtimes and dependencies across team members and deployment environments. They can eliminate environment-specific bugs and reduce onboarding time for new team members.

To use a development container, start by creating a `.devcontainer` folder in your repository root. Next, create a `devcontainer.json`, `Dockerfile`, and `requirements.txt` file.

```json  theme={null}
// .devcontainer/devcontainer.json
{
  "name": "ds-env",
  "build": { "dockerfile": "Dockerfile" },
  "features": {
    "ghcr.io/devcontainers/features/python:1": { "version": "3.11" }
  },
  "postCreateCommand": "pip install -r requirements.txt"
}
```

```dockerfile  theme={null}
# .devcontainer/Dockerfile
FROM mcr.microsoft.com/devcontainers/python:3.11
COPY requirements.txt .
RUN pip install --upgrade pip && \
    pip install -r requirements.txt
```

```txt  theme={null}
# requirements.txt
pandas==2.3.0
numpy
# add other dependencies you need for your project
```

Cursor will automatically detect the devcontainer and prompt you to reopen your project within a container. Alternatively, you can manually reopen in a container using the Command Palette (<Kbd>Cmd+Shift+P</Kbd>) and searching for `Reopen in Container`.

Development containers provide several advantages:

* **Dependency isolation** prevents conflicts between projects
* **Reproducible builds** ensure consistent behavior across development and production environments
* **Simplified onboarding** allows new team members to start immediately without manual setup

### Remote development with SSH

When your analysis requires additional compute resources, GPUs, or access to private datasets, connect to remote machines while maintaining your local development environment.

1. Provision a cloud instance or access an on-premises server with required resources
2. Clone your repository to the remote machine, including the `.devcontainer` configuration
3. Connect through Cursor: <Kbd>Cmd+Shift+P</Kbd> → "Remote-SSH: Connect to Host"

This approach maintains consistent tooling while scaling compute resources as needed. The same development container configuration works across local and remote environments.

# Large Codebases
Source: https://docs.cursor.com/en/guides/advanced/large-codebases

How to work with large codebases in Cursor

Working with large codebases introduces a new set of challenges than working on smaller projects. Drawing from both our experience scaling Cursor's own codebase and insights from customers managing massive codebases, we've discovered some useful patterns for handling increased complexity.

In this guide, we'll walk through some of these techniques that we've found be useful for large codebases.

```mermaid  theme={null}
flowchart LR
    A[Build codebase understanding] --> B[Define outcome / Diff]
    B --> C[Plan changes]
    C --> D[Implement changes]
    D --> A
```

# Use Chat to quickly get up to speed on unfamiliar code

Navigating a large codebase, especially if it's new to you, can be challenging. You often grep, search, and click around to find the specific parts of the codebase you're looking for. With [Chat](/en/chat/overview), you can start asking questions to find what you're looking for and get a detailed explanation of how it works.

Here we're getting help to find implementation details of codebase indexing in Cursor, and even asking for some examples to make it easier to understand.

# Write rules for domain-specific knowledge

If you were onboarding a new collaborator into your codebase, what context would you give them to make sure they can start doing meaningful contributions?

Your answer to this question is likely valuable information for Cursor to understand as well. For every organization or project, there's latent knowledge that might not be fully captured in your documentation. Using rules effectively is the single best way to ensure Cursor is getting the full picture.

For example, if you're writing instructions for how to implement a new feature or service, consider writing a short rule to document it for posterity.

```mdc Boilerplate theme={null}
---
description: Add a new VSCode frontend service
---

1. **Interface Definition:**
   - Define a new service interface using `createDecorator` and ensure `_serviceBrand` is included to avoid errors.

2. **Service Implementation:**
   - Implement the service in a new TypeScript file, extending `Disposable`, and register it as a singleton with `registerSingleton`.

3. **Service Contribution:**
   - Create a contribution file to import and load the service, and register it in the main entrypoint.

4. **Context Integration:**
   - Update the context to include the new service, allowing access throughout the application.
```

If there are common formatting patterns that you want to make sure Cursor adheres to, consider auto-attaching rules based on glob patterns.

```mdc Formatting theme={null}
---
globs: *.ts
---
- Use bun as package manager. See [package.json](mdc:backend/reddit-eval-tool/package.json) for scripts
- Use kebab-case for file names
- Use camelCase for function and variable names
- Use UPPERCASE_SNAKE_CASE for hardcoded constants
- Prefer `function foo()` over `const foo = () =>`
- Use `Array<T>` instead of `T[]`
- Use named exports over default exports, e.g (`export const variable ...`, `export function `)
```

# Stay close to the plan-creation process

For larger changes, spending an above-average amount of thought to create a precise, well-scoped plan can significantly improve Cursor's output.

If you find that you're not getting the result you want after a few different variations of the same prompt, consider zooming out and creating a more detailed plan from scratch, as if you were creating a PRD for a coworker. Oftentimes **the hard part is figuring out what** change should be made, a task suited well for humans. With the right instructions, we can delegate some parts of the implementation to Cursor.

One way to use AI to augment the plan-creation process is to use Ask mode. To create a plan, turn on Ask mode in Cursor and dump whatever context you have from your project management systems, internal docs, or loose thoughts. Think about what files and dependencies you have in the codebase that you already know you want to include. This can be a file that includes pieces of code you want to integrate with, or perhaps a whole folder.

Here's an example prompt:

```mdc Planning prompt theme={null}
- create a plan for how we shoud create a new feature (just like @existingfeature.ts)
- ask me questions (max 3) if anything is unclear
- make sure to search the codebase

@Past Chats (my earlier exploration prompts)

here's some more context from [project management tool]:
[pasted ticket description]
```

We're asking the model to create a plan and gather context by asking the human questions, referencing any earlier exploration prompts and also the ticket descriptions. Using a thinking model like `claude-3.7-sonnet`, `gemini-2.5-pro`, or `o3` is recommended as they can understand the intent of the change and better synthesize a plan.

From this, you can iteratively formulate the plan with the help of Cursor before starting implementation.

```mermaid  theme={null}
flowchart LR
    A[Context] -- Ask --> B[Plan]
    B -- Agent --> C[Implementation]
```

# Pick the right tool for the job

One of the most important skills in using Cursor effectively is choosing the right tool for the job. Think about what you're trying to accomplish and pick the approach that will keep you in flow.

Each tool has its sweet spot:

* Tab is your go-to for quick edits where you want to be in the driver's seat
* Inline Edit shines when you need to make focused changes to a specific section of code
* Chat is perfect for those bigger changes where you need Cursor to understand the broader context

When you're using Chat mode (which can feel a bit slower but is incredibly powerful), help it help you by providing good context. Use [@files](/en/context/@-symbols/@-files) to point to similar code you want to emulate, or [@folder](/en/context/@-symbols/@-folders) to give it a better understanding of your project structure. And don't be afraid to break bigger changes into smaller chunks - starting fresh chats helps keep things focused and efficient.

# Takeaways

* Scope down changes and don't try to do too much at once
* Include relevant context when you can
* Use Chat, Inline Edit & Tab for what they're best at
* Create new chats often
* Plan with [Ask mode](/en/chat/ask), implement with [Agent mode](/en/chat/agent)

# Working with Documentation
Source: https://docs.cursor.com/en/guides/advanced/working-with-documentation

How to leverage documentation effectively in Cursor through prompting, external sources, and internal context

export const ChatInput = ({content = []}) => {
  const renderContent = () => {
    return content.map((item, index) => {
      if (item.type === 'mention') {
        return <span key={index} className="mention bg-blue-500/20 px-1 py-0.5 rounded-sm">
                        {item.text}
                    </span>;
      }
      return item.text;
    });
  };
  return <>
            <div className="flex flex-col items-stretch border border-neutral-500 rounded-lg p-3 gap-2 bg-neutral-800 relative transition-all duration-100 ease-in-out hover:border-neutral-500">
                <div className="flex flex-col gap-1">
                    <div className="flex flex-col gap-1 outline-none overflow-hidden">
                        <div className="flex-1 flex items-center gap-2">
                            <div className="w-full box-border max-h-10 overflow-hidden">
                                <div className="flex items-center gap-2 w-full flex-nowrap">
                                    <div className="cursor-pointer flex items-center justify-center p-1 h-5 w-5 rounded border border-neutral-600 outline-none flex-shrink-0 hover:bg-neutral-700 bg-neutral-750">
                                        <span className="text-neutral-400 text-sm font-semibold">@</span>
                                    </div>
                                </div>
                            </div>
                        </div>
                    </div>
                </div>

<div className="leading-6 text-sm text-neutral-200 bg-transparent block break-words p-0 whitespace-pre-wrap font-medium min-h-6">
                                                {content.length > 0 ? renderContent() : <span className="text-neutral-500">Plan, search, build anything</span>}
                                            </div>

</div>
                                    </div>
                                </div>
                            </div>
                        </div>
                    </div>

<div className="flex items-center justify-between gap-3 flex-shrink-0 mt-6">
                        <div className="flex-1 w-full h-full flex items-center flex-col gap-1">
                            <div className="flex items-center justify-between gap-2 flex-shrink-0 w-full">
                                <div className="flex items-center justify-between w-full">
                                    <div className="flex items-center gap-3 flex-shrink min-w-0">
                                        <div className="flex gap-1 text-xs items-center min-w-0 max-w-full px-1.5 py-0.5 flex-shrink-0 cursor-pointer bg-neutral-700 hover:bg-neutral-600 rounded-full">
                                            <div className="flex items-center gap-1 min-w-0 max-w-full overflow-hidden">
                                                <div className="text-xs flex-shrink-0 w-3 h-3 flex items-center justify-center text-neutral-400">
                                                    ∞
                                                </div>
                                                <div className="min-w-0 max-w-full overflow-hidden text-ellipsis whitespace-nowrap flex items-center gap-1 font-medium">
                                                    <span className="text-neutral-300">Agent</span>
                                                    <span className="text-neutral-500 text-[10px]">⌘I</span>
                                                </div>
                                                <Icon icon="chevron-down" size={6} color="currentColor" />
                                            </div>
                                        </div>

<div className="flex items-center gap-3 justify-end">
                                        <button className="bg-white/80 border-none text-neutral-500 flex w-5 h-5 items-center justify-center hover:text-neutral-400 hover:bg-white/90 rounded-full disabled:opacity-50" disabled={content.length === 0 || !content.some(item => item.text.trim())}>
                                            <span className="text-sm">↑</span>
                                        </button>
                                    </div>
                                </div>
                            </div>
                        </div>
                    </div>
                </div>
            </div>
        </>;
};

# Why documentation matters

Documentation provides current, accurate context. Without it, models use outdated or incomplete training data. Documentation helps models understand things like:

* Current APIs and parameters
* Best practices
* Organization conventions
* Domain terminology

And much more. Read on to learn how to use documentation right in Cursor without having to context switch.

## Model knowledge cutoff

Large language models are trained on data up to a specific point in time, called a "knowledge cutoff." This means:

* Recent library updates might not be reflected
* New frameworks or tools may be unknown
* API changes after the cutoff date are missed
* Best practices may have evolved since training

For example, if a model's knowledge cutoff is early 2024, it won't know about features released in late 2024, even for popular frameworks.

# Which tool should I use?

Use this decision tree to quickly determine the best approach for your documentation needs:

```mermaid  theme={null}
flowchart TD
    A[What information do you need?] --> B[Public frameworks/libraries]
    A --> C[Recent community knowledge/troubleshooting]
    A --> D[Internal company information]
    
    B --> E[Official documentation needed?]
    E -->|Yes| F[Use @Docs<br/>API references, guides, best practices]
    E -->|No| G[Use @Web<br/>Community tutorials, comparisons]
    
    C --> H[Use @Web<br/>Recent posts, GitHub issues]
    
    D --> I[Existing MCP integrations available?]
    I -->|Yes| J[Use existing MCP<br/>Confluence, Google Drive, etc.]
    I -->|No| K[Build custom MCP<br/>Internal APIs, proprietary systems]
    
    style F fill:#e1f5fe
    style G fill:#e8f5e8  
    style H fill:#e8f5e8
    style J fill:#fff3e0
    style K fill:#fce4ec
```

## Mental model

<div className="full-width-table">
  | Tool        | Mental Model                                     |
  | ----------- | ------------------------------------------------ |
  | **`@Docs`** | Like browsing and reading official documentation |
  | **`@Web`**  | Like searching for solutions on the internet     |
  | **MCP**     | Like accessing your internal documentation       |
</div>

# Public documentation

External documentation covers publicly available information that models might have limited or outdated knowledge about. Cursor provides two primary ways to access this information.

## Using @Docs

`@Docs` connects Cursor to official documentation from popular tools and frameworks. Use it when you need current, authoritative information about:

* **API references**: Function signatures, parameters, return types
* **Getting started guides**: Setup, configuration, basic usage
* **Best practices**: Recommended patterns from the source
* **Framework-specific debugging**: Official troubleshooting guides

## Using @Web

`@Web` searches the live internet for current information, blog posts, and community discussions. Use it when you need:

* **Recent tutorials**: Community-generated content and examples
* **Comparisons**: Articles comparing different approaches
* **Recent updates**: Very recent updates or announcements
* **Multiple perspectives**: Different approaches to problems

# Internal documentation

Internal documentation includes information specific to your organization that AI models have never encountered during training. This might be:

* **Internal APIs**: Custom services and microservices
* **Company standards**: Coding conventions, architecture patterns
* **Proprietary systems**: Custom tools, databases, workflows
* **Domain knowledge**: Business logic, compliance requirements

## Accessing internal docs with MCP

Model Context Protocol (MCP) provides a standardized way to bring your private documentation and systems into Cursor. MCP acts as a thin layer between Cursor and your internal resources.

**Why MCP matters:**

* Models can't guess your internal conventions
* API documentation for custom services isn't publicly available
* Business logic and domain knowledge is unique to your organization
* Compliance and security requirements vary by company

### Common MCP integrations

| Integration      | Access                         | Examples                                                                                                                     |
| ---------------- | ------------------------------ | ---------------------------------------------------------------------------------------------------------------------------- |
| **Confluence**   | Company Confluence spaces      | Architecture documentation, API specifications for internal services, coding standards and guidelines, process documentation |
| **Google Drive** | Shared documents and folders   | Specification documents, meeting notes and decision records, design documents and requirements, team knowledge bases         |
| **Notion**       | Workspace databases and pages  | Project documentation, team wikis, knowledge bases, product requirements, technical specifications                           |
| **Custom**       | Internal systems and databases | Proprietary APIs, legacy documentation systems, custom knowledge bases, specialized tools and workflows                      |

#### Custom solutions

For unique needs, you can build custom MCP servers that:

* Scrape internal websites or portals
* Connect to proprietary databases
* Access custom documentation systems
* Pull from internal wikis or knowledge bases

<Tip>If you build a custom MCP server, you can also expose tools for Cursor to update the documentation</Tip>

Example custom MCP server for scraping internal docs:

<CodeGroup>
  ```javascript TypeScript theme={null}
  import { McpServer, ResourceTemplate } from "@modelcontextprotocol/sdk/server/mcp.js";
  import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
  import { z } from "zod";
  import TurndownService from "turndown";

// Create an MCP server for scraping internal docs
  const server = new McpServer({
    name: "internal-docs",
    version: "1.0.0"
  });

const turndownService = new TurndownService();

// Add tool to scrape internal documentation
  server.tool("get_doc",
    { url: z.string() },
    async ({ url }) => {
      try {
        const response = await fetch(url);
        const html = await response.text();
        
        // Convert HTML to markdown
        const markdown = turndownService.turndown(html);
        
        return {
          content: [{ type: "text", text: markdown }]
        };
      } catch (error) {
        return {
          content: [{ type: "text", text: `Error scraping ${url}: ${error.message}` }]
        };
      }
    }
  );

// Start receiving messages on stdin and sending messages on stdout
  const transport = new StdioServerTransport();
  await server.connect(transport);
  ```

```python Python theme={null}
  # server.py
  import os
  import asyncio
  from mcp.server.fastmcp import FastMCP
  import aiohttp
  from markdownify import markdownify as md

# Create an MCP server for scraping internal docs
  mcp = FastMCP("internal-docs")

@mcp.tool()
  async def get_doc(url: str) -> dict:
      """Scrape internal documentation from a URL"""
      try:
          async with aiohttp.ClientSession() as session:
              async with session.get(url) as response:
                  html = await response.text()
          
          # Convert HTML to markdown
          markdown = md(html)
          
          return {
              "content": [{"type": "text", "text": markdown}]
          }
      except Exception as error:
          return {
              "content": [{"type": "text", "text": f"Error scraping {url}: {str(error)}"}]
          }
  ```
</CodeGroup>

# Keeping docs up to date

Documentation becomes stale quickly. Cursor can help you maintain current, useful documentation by generating and updating it based on your actual code and development conversations.

## From existing code

Use Cursor to create documentation directly from your codebase:

## From chat sessions

Your conversations with Cursor contain valuable intent that can be turned into documentation.

<Tabs>
  <Tab title="Problem Solving">
    **After solving a complex problem:**

<Tab title="Architecture">
    **After architectural decisions:**

<Tab title="Debugging">
    **After debugging sessions:**

## Takeaways

* Documentation as context makes Cursor more accurate and current
* Use `@Docs` for official documentation and `@Web` for community knowledge
* MCP bridges the gap between Cursor and your internal systems
* Generate documentation from code and conversations to keep knowledge current
* Combine external and internal documentation sources for comprehensive understanding

# Java
Source: https://docs.cursor.com/en/guides/languages/java

Set up Java development with JDK, extensions, and build tools

This guide will help you configure Cursor for Java development, including setting up the JDK, installing necessary extensions, debugging, running Java applications, and integrating build tools like Maven and Gradle. It also covers workflow features similar to IntelliJ or VS Code.

<Note>
  Before starting, ensure you have Cursor installed and updated to the latest
  version.
</Note>

## Setting up Java for Cursor

### Java Installation

Before setting up Cursor itself, you will need Java installed on your machine.

<Warning>
  Cursor does not ship with a Java compiler, so you need to install a JDK if you
  haven't already.
</Warning>

<CardGroup cols={1}>
  <Card title="Windows Installation" icon="windows">
    Download and install a JDK (e.g., OpenJDK, Oracle JDK, Microsoft Build of
    OpenJDK).

<br />

Set JAVA\_HOME and add JAVA\_HOME\bin to your PATH.
  </Card>

<Card title="macOS Installation" icon="apple">
    Install via Homebrew (`brew install openjdk`) or download an installer.

<br />

Ensure JAVA\_HOME points to the installed JDK.
  </Card>

<Card title="Linux Installation" icon="linux">
    Use your package manager (`sudo apt install openjdk-17-jdk` or equivalent)
    or install via SDKMAN.
  </Card>
</CardGroup>

To check installation, run:

```bash  theme={null}
java -version
javac -version
```

<Info>
  If Cursor does not detect your JDK, configure it manually in settings.json:
</Info>

```json  theme={null}
{
  "java.jdt.ls.java.home": "/path/to/jdk",
  "java.configuration.runtimes": [
    {
      "name": "JavaSE-17",
      "path": "/path/to/jdk-17",
      "default": true
    }
  ]
}
```

<Warning>Restart Cursor to apply changes.</Warning>

### Cursor Setup

<Info>Cursor supports VS Code extensions. Install the following manually:</Info>

<CardGroup cols={2}>
  <Card title="Extension Pack for Java" icon="java" href="cursor:extension/vscjava.vscode-java-pack">
    Includes Java language support, debugger, test runner, Maven support, and
    project manager
  </Card>

<Card title="Gradle for Java" icon="gears" href="cursor:extension/vscjava.vscode-gradle">
    Essential for working with Gradle build system
  </Card>

<Card title="Spring Boot Extension Pack" icon="leaf" href="cursor:extension/vmware.vscode-boot-dev-pack">
    Required for Spring Boot development
  </Card>

<Card title="Kotlin" icon="window" href="cursor:extension/fwcd.kotlin">
    Necessary for Kotlin application development
  </Card>
</CardGroup>

### Configure Build Tools

#### Maven

Ensure Maven is installed (`mvn -version`). Install from [maven.apache.org](https://maven.apache.org/download.cgi) if needed:

1. Download the binary archive
2. Extract to desired location
3. Set MAVEN\_HOME environment variable to the extracted folder
4. Add %MAVEN\_HOME%\bin (Windows) or \$MAVEN\_HOME/bin (Unix) to PATH

#### Gradle

Ensure Gradle is installed (`gradle -version`). Install from [gradle.org](https://gradle.org/install/) if needed:

1. Download the binary distribution
2. Extract to desired location
3. Set GRADLE\_HOME environment variable to the extracted folder
4. Add %GRADLE\_HOME%\bin (Windows) or \$GRADLE\_HOME/bin (Unix) to PATH

Alternatively, use the Gradle Wrapper which will automatically download and use the correct Gradle version:

## Running and Debugging

Now you are all set up, it's time to run and debug your Java code.
Depending on your needs, you can use the following methods:

<CardGroup cols={2}>
  <Card title="Run" icon="play">
    Click the "Run" link that appears above any main method to quickly execute
    your program
  </Card>

<Card title="Debug" icon="bug">
    Open the Run and Debug sidebar panel and use the Run button to start your
    application
  </Card>
</CardGroup>

<CardGroup cols={1}>
  <Card title="Terminal" icon="terminal">
    Execute from command line using Maven or Gradlecommands
  </Card>

<Card title="Spring Boot" icon="leaf">
    Launch Spring Boot applications directly from the Spring Boot Dashboard
    extension
  </Card>
</CardGroup>

## Java x Cursor Workflow

Cursor's AI-powered features can significantly enhance your Java development workflow. Here are some ways to leverage Cursor's capabilities specifically for Java:

<CardGroup cols={2}>
  <Card title="Tab Completion" icon="arrow-right">
    <div className="text-sm">
      Smart completions for methods, signatures, and Java boilerplate like
      getters/setters.
    </div>
  </Card>

<Card title="Agent Mode" icon="pen-to-square">
    <div className="text-sm">
      Implement design patterns, refactor code, or generate classes with proper
      inheritance.
    </div>
  </Card>

<Card title="Inline Edit" icon="code">
    <div className="text-sm">
      Quick inline edits to methods, fix errors, or generate unit tests without
      breaking flow.
    </div>
  </Card>

<Card title="Chat" icon="message">
    <div className="text-sm">
      Get help with Java concepts, debug exceptions, or understand framework
      features.
    </div>
  </Card>
</CardGroup>

### Example Workflows

1. **Generate Java Boilerplate**\
   Use [Tab completion](/en/tab/overview) to quickly generate constructors, getters/setters, equals/hashCode methods, and other repetitive Java patterns.

2. **Debug Complex Java Exceptions**\
   When facing a cryptic Java stack trace, highlight it and use [Ask](/en/chat/overview) to explain the root cause and suggest potential fixes.

3. **Refactor Legacy Java Code**\
   Use [Agent mode](/en/chat/agent) to modernize older Java code - convert anonymous classes to lambdas, upgrade to newer Java language features, or implement design patterns.

4. **Frameworks Development**\
   Add your documentation to Cursor's context with @docs, and generate framework-specific code throughout Cursor.

# JavaScript & TypeScript
Source: https://docs.cursor.com/en/guides/languages/javascript

JavaScript and TypeScript development with framework support

Welcome to JavaScript and TypeScript development in Cursor! The editor provides exceptional support for JS/TS development through its extension ecosystem. Here's what you need to know to get the most out of Cursor.

## Essential Extensions

While Cursor works great with any extensions you prefer, we recommend these for those just getting started:

* **ESLint** - Required for Cursor's AI-powered lint fixing capabilities
* **JavaScript and TypeScript Language Features** - Enhanced language support and IntelliSense
* **Path Intellisense** - Intelligent path completion for file paths

## Cursor Features

Cursor enhances your existing JavaScript/TypeScript workflow with:

* **Tab Completions**: Context-aware code completions that understand your project structure
* **Automatic Imports**: Tab can automatically import libraries as soon as you use them
* **Inline Editing**: Use `CMD+K` on any line to edit with perfect syntax
* **Composer Guidance**: Plan and edit your code across multiple files with the Composer

### Framework Intelligence with @Docs

Cursor's @Docs feature lets you supercharge your JavaScript development by adding custom documentation sources that the AI can reference. Add documentation from MDN, Node.js, or your favorite framework to get more accurate and contextual code suggestions.

<Card title="Learn more about @Docs" icon="book" href="/en/context/@-symbols/@-docs">
  Discover how to add and manage custom documentation sources in Cursor.
</Card>

### Automatic Linting Resolution

One of Cursor's standout features is its seamless integration with Linter extensions.
Ensure you have a linter, like ESLint, setup, and enable the 'Iterate on Lints' setting.

Then, when using the Agent mode in Composer, once the AI has attempted to answer your query, and has made any code changes, it will automatically read the output of the linter and will attempt to fix any lint errors it might not have known about.

## Framework Support

Cursor works seamlessly with all major JavaScript frameworks and libraries, such as:

### React & Next.js

* Full JSX/TSX support with intelligent component suggestions
* Server component and API route intelligence for Next.js
* Recommended: [**React Developer Tools**](cursor:extension/msjsdiag.vscode-react-native) extension

### Vue.js

* Template syntax support with Volar integration
* Component auto-completion and type checking
* Recommended: [**Vue Language Features**](cursor:extension/vue.volar)

### Angular

* Template validation and TypeScript decorator support
* Component and service generation
* Recommended: [**Angular Language Service**](cursor:extension/Angular.ng-template)

### Svelte

* Component syntax highlighting and intelligent completions
* Reactive statement and store suggestions
* Recommended: [**Svelte for VS Code**](cursor:extension/svelte.svelte-vscode)

### Backend Frameworks (Express/NestJS)

* Route and middleware intelligence
* TypeScript decorator support for NestJS
* API testing tools integration

Remember, Cursor's AI features work well with all these frameworks, understanding their patterns and best practices to provide relevant suggestions. The AI can help with everything from component creation to complex refactoring tasks, while respecting your project's existing patterns.

# Python
Source: https://docs.cursor.com/en/guides/languages/python

Set up Python development with extensions and linting tools

<Note>
  This guide was heavily inspired by [Jack Fields](https://x.com/OrdinaryInds)
  and his
  [article](https://medium.com/ordinaryindustries/the-ultimate-vs-code-setup-for-python-538026b34d94)
  about setting up VS Code for Python development. Please check his article for
  more details.
</Note>

## Prerequisites

Before we begin, ensure you have:

* [Python](https://python.org) installed (3.8 or higher recommended)
* [Git](https://git-scm.com/) for version control
* Cursor installed and updated to the latest version

## Essential Extensions

The following extensions setup Cursor to be fully featured for Python development. These provide you with syntax highlighting, linting, debugging and unit testing.

<CardGroup cols={2}>
  <Card title="Python" icon="python" href="cursor:extension/ms-python.python">
    Core language support from Microsoft
  </Card>

<Card title="Cursor Pyright" icon="bolt" href="cursor:extension/anysphere.cursorpyright">
    Fast Python language server
  </Card>

<Card title="Python Debugger" icon="bug" href="cursor:extension/ms-python.debugpy">
    Enhanced debugging capabilities
  </Card>

<Card title="Ruff" icon="wand-magic-sparkles" href="cursor:extension/charliermarsh.ruff">
    Python linter and formatter
  </Card>
</CardGroup>

### Advanced Python Tooling

While the above extensions have previously been the most popular extensions for Python development in Cursor, we've also added some additional extensions that can help you get the most out of your Python development.

#### `uv` - Python Environment Manager

[uv](https://github.com/astral-sh/uv) is a modern Python package manager that can be used to create and manage virtual environments, in addition to replacing pip as the default package manager.

To install uv, run the following command in your terminal:

```bash  theme={null}
pip install uv
```

#### `ruff` - Python Linter and Formatter

[Ruff](https://docs.astral.sh/ruff/) is a modern Python linter and formatter that can be used to check for programming errors, helps enforce coding standards, and can suggest refactoring. It can be used alongside Black for code formatting.

To install Ruff, run the following command in your terminal:

```bash  theme={null}
pip install ruff
```

## Cursor Configuration

### 1. Python Interpreter

Configure your Python interpreter in Cursor:

1. Open Command Palette (Cmd/Ctrl + Shift + P)
2. Search for "Python: Select Interpreter"
3. Choose your Python interpreter (or virtual environment if you're using one)

### 2. Code Formatting

Set up automatic code formatting with Black:

<Note>
  Black is a code formatter that automatically formats your code to follow a
  consistent style. It requires zero configuration and is widely adopted in the
  Python community.
</Note>

To install Black, run the following command in your terminal:

```bash  theme={null}
pip install black
```

Then, configure Cursor to use Black for code formatting, by adding the following to your `settings.json` file:

```json  theme={null}
{
  "python.formatting.provider": "black",
  "editor.formatOnSave": true,
  "python.formatting.blackArgs": ["--line-length", "88"]
}
```

### 3. Linting

We can use PyLint to check for programming errors, helps enforce coding standards, and can suggest refactoring.

To install PyLint, run the following command in your terminal:

```bash  theme={null}
pip install pylint
```

```json  theme={null}
{
  "python.linting.enabled": true,
  "python.linting.pylintEnabled": true,
  "python.linting.lintOnSave": true
}
```

### 4. Type Checking

In addition to linting, we can use MyPy to check for type errors.

To install MyPy, run the following command in your terminal:

```bash  theme={null}
pip install mypy
```

```json  theme={null}
{
  "python.linting.mypyEnabled": true
}
```

## Debugging

Cursor provides powerful debugging capabilities for Python:

1. Set breakpoints by clicking the gutter
2. Use the Debug panel (Cmd/Ctrl + Shift + D)
3. Configure `launch.json` for custom debug configurations

## Recommended Features

<CardGroup cols={3}>
  <Card title="Tab Completion" icon="wand-magic-sparkles" href="/en/tab/overview">
    Intelligent code suggestions that understand your actions
  </Card>

<Card title="Chat" icon="comments" href="/en/chat/overview">
    Explore and understand code through natural conversations
  </Card>

<Card title="Agent" icon="robot" href="/en/chat/agent">
    Handle complex development tasks with AI assistance
  </Card>

<Card title="Context" icon="network-wired" href="/en/context/model-context-protocol">
    Pull in context from 3rd party systems
  </Card>

<Card title="Auto-Imports" icon="file-import" href="/en/tab/auto-import">
    Automatically import modules as you code
  </Card>

<Card title="AI Review" icon="check-double" href="/en/tab/overview#quality">
    Cursor constantly reviews your code with AI
  </Card>
</CardGroup>

## Framework Support

Cursor works seamlessly with popular Python frameworks:

* **Web Frameworks**: Django, Flask, FastAPI
* **Data Science**: Jupyter, NumPy, Pandas
* **Machine Learning**: TensorFlow, PyTorch, scikit-learn
* **Testing**: pytest, unittest
* **API**: requests, aiohttp
* **Database**: SQLAlchemy, psycopg2

# iOS & macOS (Swift)
Source: https://docs.cursor.com/en/guides/languages/swift

Integrate Cursor with Xcode for Swift development

Welcome to Swift development in Cursor! Whether you're building iOS apps, macOS applications, or server-side Swift projects, we've got you covered. This guide will help you set up your Swift environment in Cursor, starting with the basics and moving on to more advanced features.

## Basic Workflow

The simplest way to use Cursor with Swift is to treat it as your primary code editor while still relying on Xcode for building and running your apps. You'll get great features like:

* Smart code completion
* AI-powered coding assistance (try [CMD+K](/en/inline-edit/overview) on any line)
* Quick access to documentation with [@Docs](/en/context/@-symbols/@-docs)
* Syntax highlighting
* Basic code navigation

When you need to build or run your app, simply switch to Xcode. This workflow is perfect for developers who want to leverage Cursor's AI capabilities while sticking to familiar Xcode tools for debugging and deployment.

### Hot Reloading

When using Xcode workspaces or projects (instead of opening a folder directly in Xcode), Xcode can often ignore changes to your files that were made in Cursor, or outside of Xcode in general.

While you can open the folder in Xcode to resolve this, you may need to use a project for your Swift development workflow.

A great solution to this is to use [Inject](https://github.com/krzysztofzablocki/Inject), a hot reloading library for Swift that allows your app to "hot reload" and update as soon as changes are made in real time. This does not suffer from the side effects of the Xcode workspace/project issue, and allows you to make changes in Cursor and have them reflected in your app immediately.

<CardGroup cols={1}>
  <Card title="Inject - Hot Reloading for Swift" horizontal icon="fire" href="https://github.com/krzysztofzablocki/Inject">
    Learn more about Inject and how to use it in your Swift projects.
  </Card>
</CardGroup>

## Advanced Swift Development

<Note>
  This section of the guide was heavily inspired by [Thomas
  Ricouard](https://x.com/Dimillian) and his
  [article](https://dimillian.medium.com/how-to-use-cursor-for-ios-development-54b912c23941)
  about using Cursor for iOS development. Please check his article for more
  details and drop him a follow for more Swift content.
</Note>

If you are looking to only need one editor open at a time, and want to avoid the need to switch between Xcode and Cursor, you can use an extension like [Sweetpad](https://sweetpad.hyzyla.dev/) to integrate Cursor directly with Xcode's underlying build system.

Sweetpad is a powerful extension that allows you to build, run and debug your Swift projects directly in Cursor, without compromising on Xcode's features.

To get started with Sweetpad, you'll still need to have Xcode installed on your Mac - it's the foundation of Swift development. You can download Xcode from the [Mac App Store](https://apps.apple.com/us/app/xcode/id497799835). Once you have Xcode set up, let's enhance your development experience in Cursor with a few essential tools.

Open your terminal and run:

```bash  theme={null}
# Builds your projects without needing Xcode open
brew install xcode-build-server

# Pretty print's the `xcodebuild` command output into Cursor's terminal
brew install xcbeautify

# Allows for advanced formating and language features
brew install swiftformat
```

Next, install the [Swift Language Support](cursor:extension/chrisatwindsurf.swift-vscode) extension in Cursor. This will give you syntax highlighting and basic language features right out of the box.

Then, we can install the [Sweetpad](https://sweetpad.hyzyla.dev/) extension to integrate Cursor with Xcode. Sweetpad wraps a bunch of shortcuts around the `xcodebuild` CLI (and much more), and allows you to scan your targets, select the destination, build, and run your app just like Xcode. On top of that, it’ll set up your project for Xcode Build Server so you get all the features mentioned above.

### Sweetpad Usage

Once Sweetpad is installed, and you have a Swift project open in Cursor, you should first run the `Sweetpad: Generate Build Server Config` command. This will generate a `buildServer.json` file in the root of your project that allows the Xcode Build Server to work with your project.

Then, from either the Command Palette or the Sweetpad sidebar, you can select the target you want to build and run.

<Note>
  {" "}

You need to build your project once to enable auto-completion, jump to definition,
  and other language features.{" "}
</Note>

You can also now hit F5 to build and run your project with a debugger - you might need to create a launch configuration first, but just select Sweetpad from the list when prompted!

As with many extensions in Cursor, you can bind many of the Sweetpad commands to keyboard shortcuts, to make your workflow even more efficient.

To learn more about Sweetpad, check out these resources:

<CardGroup>
  <Card title="Sweetpad Website" horizontal icon="globe" href="https://sweetpad.hyzyla.dev/">
    Official Sweetpad website with features and installation instructions
  </Card>

<Card title="Sweetpad Guide" horizontal icon="book" href="https://sweetpad.hyzyla.dev/docs/intro">
    Comprehensive guide covering configuration, usage and advanced features
  </Card>
</CardGroup>

# JetBrains
Source: https://docs.cursor.com/en/guides/migration/jetbrains

Migrate from JetBrains IDEs to Cursor with familiar tools

Cursor offers a modern, AI-powered coding experience that can replace your JetBrains IDEs. While the transition might feel different at first, Cursor's VS Code-based foundation provides powerful features and extensive customization options.

## Editor Components

### Extensions

JetBrains IDEs are great tools, as they come already pre-configured for the languages and frameworks they are intended for.

Cursor is different - being a blank canvas out of the box, you can customize it to your liking, not being limited by the languages and frameworks the IDE was intended for.

Cursor has access to a vast ecosystem of extensions, and almost all of the functionality (and more!) that JetBrains IDEs offer can be recreated through these extensions.

Take a look at some of these popular extensions below:

<CardGroup cols={4}>
  <Card title="Remote SSH" icon="network-wired" href="cursor:extension/anysphere.remote-ssh">
    SSH Extension
  </Card>

<Card title="Project Manager" icon="folder-tree" href="cursor:extension/alefragnani.project-manager">
    Manage multiple projects
  </Card>

<Card title="GitLens" icon="git" href="cursor:extension/eamodio.gitlens">
    Enhanced Git integration
  </Card>

<Card title="Local History" icon="clock-rotate-left" href="cursor:extension/xyz.local-history">
    Track local file changes
  </Card>

<Card title="Error Lens" icon="bug" href="cursor:extension/usernamehw.errorlens">
    Inline error highlighting
  </Card>

<Card title="ESLint" icon="code-compare" href="cursor:extension/dbaeumer.vscode-eslint">
    Code linting
  </Card>

<Card title="Prettier" icon="wand-magic-sparkles" href="cursor:extension/esbenp.prettier-vscode">
    Code formatting
  </Card>

<Card title="Todo Tree" icon="folder-tree" href="cursor:extension/Gruntfuggly.todo-tree">
    Track TODOs and FIXMEs
  </Card>
</CardGroup>

### Keyboard Shortcuts

Cursor has a built-in keyboard shortcut manager that allows you to map your favorite keyboard shortcuts to actions.

With this extension, you can bring almost all of the JetBrains IDEs shortcuts directly to Cursor!
Be sure to read the extension's documentation to learn how to configure it to your liking:

<Card title="IntelliJ IDEA Keybindings" icon="keyboard" href="cursor:extension/k--kato.intellij-idea-keybindings">
  Install this extension to bring JetBrains IDEs keyboard shortcuts to Cursor.
</Card>

<Note>
  Common shortcuts that differ:

* Find Action: ⌘/Ctrl+Shift+P  (vs. ⌘/Ctrl+Shift+A)
  * Quick Fix: ⌘/Ctrl+.  (vs. Alt+Enter)
  * Go to File: ⌘/Ctrl+P  (vs. ⌘/Ctrl+Shift+N)
</Note>

### Themes

Recreate the look and feel of your favorite JetBrains IDEs in Cursor with these community themes.

Choose from the standard Darcula Theme, or pick a theme to match the syntax highlighting of your JetBrains tools.

<CardGroup cols={1}>
  <Card title="JetBrains - Darcula Theme" icon="moon" horizontal href="cursor:extension/rokoroku.vscode-theme-darcula">
    Experience the classic JetBrains Darcula dark theme
  </Card>
</CardGroup>

<CardGroup cols={1}>
  <Card title="JetBrains Icons" icon="icons" horizontal href="cursor:extension/ardonplay.vscode-jetbrains-icon-theme">
    Get the familiar JetBrains file and folder icons
  </Card>
</CardGroup>

### Font

To complete your JetBrains-like experience, you can use the official JetBrains Mono font:

1. Download and install JetBrains Mono font onto your system:

2. Restart Cursor after installing the font
3. Open Settings in Cursor (⌘/Ctrl + ,)
4. Search for "Font Family"
5. Set the font family to `'JetBrains Mono'`

<Note>
  For the best experience, you can also enable font ligatures by setting `"editor.fontLigatures": true` in your settings.
</Note>

## IDE-Specific Migration

Many users loved the JetBrains IDEs for their out-the-box support for the languages and frameworks they were intended for. Cursor is different - being a blank canvas out of the box, you can customize it to your liking, not being limited by the languages and frameworks the IDE was intended for.

Cursor already has access to the extension ecosystem of VS Code, and almost all of the functionality (and more!) that JetBrains IDEs offer can be recreated through these extensions.

Take a look at the following suggested extensions for each JetBrains IDE below.

### IntelliJ IDEA (Java)

<CardGroup cols={2}>
  <Card title="Language Support for Java" icon="java" href="cursor:extension/redhat.java">
    Core Java language features
  </Card>

<Card title="Debugger for Java" icon="bug" href="cursor:extension/vscjava.vscode-java-debug">
    Java debugging support
  </Card>

<Card title="Test Runner for Java" icon="vial" href="cursor:extension/vscjava.vscode-java-test">
    Run and debug Java tests
  </Card>

<Card title="Maven for Java" icon="box" href="cursor:extension/vscjava.vscode-maven">
    Maven support
  </Card>
</CardGroup>

<CardGroup cols={1}>
  <Card title="Project Manager for Java" icon="folder-tree" href="cursor:extension/vscjava.vscode-java-dependency" horizontal>
    Project management tools
  </Card>
</CardGroup>

<Warning>
  Key differences:

* Build/Run configurations are managed through launch.json
  * Spring Boot tools available through ["Spring Boot Extension Pack"](cursor:extension/vmware.vscode-boot-dev-pack) extension
  * Gradle support via ["Gradle for Java"](cursor:extension/vscjava.vscode-gradle) extension
</Warning>

### PyCharm (Python)

<CardGroup cols={2}>
  <Card title="Python" icon="python" href="cursor:extension/ms-python.python">
    Core Python support
  </Card>

<Card title="Cursor Pyright" icon="bolt" href="cursor:extension/anysphere.cursorpyright">
    Fast type checking
  </Card>

<Card title="Jupyter" icon="notebook" href="cursor:extension/ms-toolsai.jupyter">
    Notebook support
  </Card>

<Card title="Ruff" icon="wand-magic-sparkles" href="cursor:extension/charliermarsh.ruff">
    Python formatter and linter
  </Card>
</CardGroup>

<Note>
  Key differences:

* Virtual environments managed through command palette
  * Debug configurations in launch.json
  * Requirements management through requirements.txt or Poetry
</Note>

### WebStorm (JavaScript/TypeScript)

<CardGroup cols={2}>
  <Card title="JavaScript and TypeScript Nightly" icon="js" href="cursor:extension/ms-vscode.vscode-typescript-next">
    Latest language features
  </Card>

<Card title="ES7+ React/Redux Snippets" icon="react" href="cursor:extension/dsznajder.es7-react-js-snippets">
    React development
  </Card>

<Card title="Vue Language Features" icon="vuejs" href="cursor:extension/Vue.volar">
    Vue.js support
  </Card>

<Card title="Angular Language Service" icon="angular" href="cursor:extension/Angular.ng-template">
    Angular development
  </Card>
</CardGroup>

<Info>
  Most WebStorm features are built into Cursor/VS Code, including:

* npm scripts view
  * Debugging
  * Git integration
  * TypeScript support
</Info>

### PhpStorm (PHP)

<CardGroup cols={2}>
  <Card title="PHP Intelephense" icon="php" href="cursor:extension/bmewburn.vscode-intelephense-client">
    PHP language server
  </Card>

<Card title="PHP Debug" icon="bug" href="cursor:extension/xdebug.php-debug">
    Xdebug integration
  </Card>

<Card title="PHP Intellisense" icon="brain" href="cursor:extension/felixfbecker.php-intellisense">
    Code intelligence
  </Card>

<Card title="PHP DocBlocker" icon="comment-dots" href="cursor:extension/neilbrayfield.php-docblocker">
    Documentation tools
  </Card>
</CardGroup>

<Note>
  Key differences:

* Xdebug configuration through launch.json
  * Composer integration via terminal
  * Database tools through ["SQLTools"](cursor:extension/mtxr.sqltools) extension
</Note>

### Rider (.NET)

<CardGroup cols={2}>
  <Card title="C#" icon="code" href="cursor:extension/anysphere.csharp">
    Core C# support
  </Card>

<Card title="DotRush" icon="toolbox" href="cursor:extension/nromanov.dotrush">
    Open-Source C# Development Environment
  </Card>

<Card title="ReSharper Plugin" icon="box" href="https://www.jetbrains.com/help/resharper-vscode/Get_started.html#installation">
    JetBrains C# Plugin
  </Card>

<Card title=".NET Install Tool" icon="box-open" href="cursor:extension/ms-dotnettools.vscode-dotnet-runtime">
    .NET SDK management
  </Card>
</CardGroup>

<Warning>
  Key differences:

* Solution explorer through file explorer
  * NuGet package management through CLI or extensions
  * Test runner integration through test explorer
</Warning>

### GoLand (Go)

<CardGroup cols={1}>
  <Card title="Go" icon="golang" href="cursor:extension/golang.Go">
    Official Go extension
  </Card>
</CardGroup>

<Note>
  Key differences:

* Go tools installation prompted automatically
  * Debugging through launch.json
  * Package management integrated with go.mod
</Note>

## Tips for a Smooth Transition

<Steps>
  <Step title="Use Command Palette">
    Press <kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd> to find commands
  </Step>

<Step title="AI Features">
    Leverage Cursor's AI features for code completion and refactoring
  </Step>

<Step title="Customize Settings">
    Fine-tune your settings.json for optimal workflow
  </Step>

<Step title="Terminal Integration">
    Use the built-in terminal for command-line operations
  </Step>

<Step title="Extensions">
    Browse the VS Code marketplace for additional tools
  </Step>
</Steps>

<Info>
  Remember that while some workflows might be different, Cursor offers powerful AI-assisted coding features that can enhance your productivity beyond traditional IDE capabilities.
</Info>

# VS Code
Source: https://docs.cursor.com/en/guides/migration/vscode

Import VS Code settings and extensions with one click

Cursor is based upon the VS Code codebase, allowing us to focus on making the best AI-powered coding experience while maintaining a familiar editing environment. This makes it easy to migrate your existing VS Code settings to Cursor.

## Profile Migration

### One-click Import

Here's how to get your entire VS Code setup in one click:

1. Open the Cursor Settings (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>J</kbd>)
2. Navigate to General > Account
3. Under "VS Code Import", click the Import button

This will transfer your:

* Extensions
* Themes
* Settings
* Keybindings

### Manual Profile Migration

If you are moving between machines, or want more control over your settings, you can manually migrate your profile.

#### Exporting a Profile

1. On your VS Code instance, open the Command Palette (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd>)
2. Search for "Preferences: Open Profiles (UI)"
3. Find the profile you want to export on the left sidebar
4. Click the 3-dot menu and select "Export Profile"
5. Choose to export it either to your local machine or to a GitHub Gist

#### Importing a Profile

1. On your Cursor instance, open the Command Palette (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd>)
2. Search for "Preferences: Open Profiles (UI)"
3. Click the dropdown menu next to 'New Profile' and click 'Import Profile'
4. Either paste in the URL of the GitHub Gist or choose 'Select File' to upload a local file
5. Click 'Import' at the bottom of the dialog to save the profile
6. Finally, in the sidebar, choose the new profile and click the tick icon to active it

## Settings and Interface

### Settings Menus

<CardGroup>
  <Card title="Cursor Settings" icon="gear">
    Access via Command Palette (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd>{" "}

* <kbd>P</kbd>), then type "Cursor Settings"
  </Card>

<Card title="VS Code Settings" icon="code">
    Access via Command Palette (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd>{" "}

* <kbd>P</kbd>), then type "Preferences: Open Settings (UI)"
  </Card>
</CardGroup>

### Version Updates

<Card title="Version Updates" icon="code-merge">
  We regularly rebase Cursor onto the latest VS Code version to stay current
  with features and fixes. To ensure stability, Cursor often uses slightly older
  VS Code versions.
</Card>

### Activity Bar Orientation

We made it horizontal to optimize space for the AI chat interface. If you prefer vertical:

1. Open the Command Palette (<kbd>⌘</kbd>/<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd>)
2. Search for "Preferences: Open Settings (UI)"
3. Search for `workbench.activityBar.orientation`
4. Set the value to `vertical`
5. Restart Cursor

# Architectural Diagrams
Source: https://docs.cursor.com/en/guides/tutorials/architectural-diagrams

Learn to generate architectural diagrams using Mermaid to visualize system structure and data flow

Architectural diagrams help you make sense of how your system works. You can use them to explore logic, track data, and communicate structure. Cursor supports generating these diagrams directly using tools like Mermaid, so you can go from code to visual in just a few prompts.

## Why diagrams matter

Diagrams clarify how data flows and how components interact. They're useful when you:

* Want to understand flow control in your codebase
* Need to trace data lineage from input to output
* Are onboarding others or documenting your system

They're also great for debugging and asking smarter questions. Visuals help you (and the model) see the bigger picture.

## Two dimensions to consider

There are a few different angles to think about:

* **Purpose**: Are you mapping logic, data flow, infrastructure, or something else?
* **Format**: Do you want something quick (like a Mermaid diagram) or formal (like UML)?

## How to prompt

Start with a clear goal. Here are some common ways to ask:

* **Flow control**: "Show me how requests go from the controller to the database."
* **Data lineage**: "Trace this variable from where it enters to where it ends up."
* **Structure**: "Give me a component-level view of this service."

You can include start and end points, or ask Cursor to find the full path.

## Working with Mermaid

Mermaid is simple to learn and renders directly in Markdown (with the right extension). Cursor can generate diagrams like:

* `flowchart` for logic and sequences
* `sequenceDiagram` for interactions
* `classDiagram` for object structure
* `graph TD` for simple directional maps

```mermaid  theme={null}
sequenceDiagram
    participant User
    participant Server
    participant Database

User->>Server: Submit Form
    Server->>Database: Save Entry
    Database-->>Server: Success
    Server-->>User: Confirmation

```

You can install the [Mermaid extension](https://marketplace.cursorapi.com/items?itemName=bierner.markdown-mermaid) to preview diagrams.

1. Go to Extensions tab
2. Search for Mermaid
3. Install

## Diagram strategy

Start small. Don't aim to map everything at once.

* Pick one function, route, or process
* Ask Cursor to diagram that part using Mermaid
* Once you have a few, ask it to combine them

This mirrors the **C4 model** – where you start at a low level (code or components) and work upward to higher-level overviews.

### Recommended flow

1. Start with a detailed, low-level diagram
2. Summarize it into a mid-level view
3. Repeat until you reach the level of abstraction you want
4. Ask Cursor to merge them into a single diagram or system map

```mermaid  theme={null}
graph TD
    subgraph Level 1: Low-level components
        A1[AuthService] --> A2[TokenValidator]
        A1 --> A3[UserDB]
        B1[PaymentService] --> B2[BillingEngine]
        B1 --> B3[InvoiceDB]
    end

subgraph Level 2: Mid-level systems
        A[User System] --> A1
        B[Billing System] --> B1
    end

subgraph Level 3: High-level application
        App[Main App] --> A
        App --> B
    end

```

## Takeaways

* Use diagrams to understand flow, logic, and data
* Start with small prompts and grow your diagram from there
* Mermaid is the easiest format to work with in Cursor
* Start low-level and abstract upward, just like in the C4 model
* Cursor can help you generate, refine, and combine diagrams with ease

# Building an MCP Server
Source: https://docs.cursor.com/en/guides/tutorials/building-mcp-server

Learn how to build a Model Context Protocol (MCP) server with PostgreSQL integration for Cursor

## Introduction

MCP servers let you connect custom data sources and make them available for use inside Cursor. This is particularly useful when you need context from places such as browsers, databases, or error and system logs. Setting up an MCP server is straightforward, and with Cursor it can be done quickly.

In this guide, we will walk through how to build an MCP server for Postgres. Our goal is to enable Cursor to run SQL queries directly against a Postgres database and to expose table schemas in a structured way.

<Note>
  This tutorial is designed to teach the fundamentals of building MCP servers.
</Note>

```mermaid  theme={null}
graph LR
    A[User in Cursor] --> B[Cursor]
    B <--> C["MCP Server (stdio)"]
    C <--> D[(Postgres Database)]

subgraph Local Machine
      B
      C
    end
    class B cursor
    class C mcp
    class D db
```

## What is an MCP Server?

An [MCP server](/en/context/mcp) is a process that communicates with Cursor and provides access to external data or actions. It can be implemented in several ways, but here we will use the simplest method: a server that runs locally on your computer over [stdio](https://en.wikipedia.org/wiki/Standard_streams) (standard input/output streams). This avoids complicated security considerations and allows us to focus on the MCP logic itself.

One of the most common use cases for MCP is database access. When building dashboards, running analyses, or creating migrations, it is often necessary to query and inspect a database. Our Postgres MCP server will support two core capabilities: running arbitrary queries and listing table schemas.

Although both of these tasks could be performed with plain SQL, MCP offers features that make them more powerful and more generally useful. Tools provide a way to expose actions such as executing queries, while resources allow us to share standardized context such as schema information. Later in this guide we will also look at prompts, which enable more advanced workflows.

Under the hood, we will rely on the postgres npm package to execute SQL statements against the database. The MCP SDK will serve as a wrapper around these calls, letting us integrate Postgres functionality seamlessly into Cursor.

## How to Build the MCP Server

The first step in building the server is to setup a new project. We’ll start with creating a new folder and initializing a Bun project

```bash  theme={null}
> mkdir postgres-mcp-server
> Bun init
```

From here, we’ll select `Blank` project. Once our boilerplate is setup, we need to install the required dependencies. `zod` is required to define schemas for i/o in MCP sdk

```bash  theme={null}
bun add postgres @modelcontextprotocol/sdk zod
```

From here, we’ll go to the repositories for each of the libraries and get the link to the raw file contents of each respective README files. We’ll use these for context when building the server

* `postgres`
  * Repo:  [https://github.com/porsager/postgres](https://github.com/porsager/postgres),
  * README: [https://raw.githubusercontent.com/porsager/postgres/refs/heads/master/README.md](https://raw.githubusercontent.com/porsager/postgres/refs/heads/master/README.md)
* `@modelcontextprotocol/sdk`:
  * Repo: [https://github.com/modelcontextprotocol/typescript-sdk](https://github.com/modelcontextprotocol/typescript-sdk)
  * README: [https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/refs/heads/main/README.md](https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/refs/heads/main/README.md)

Now, we’ll define how we want the server to behave. To do that, we’ll create a `spec.md` and write out the high level goals

```markdown  theme={null}
# Spec

- Allow defining DATABASE_URL through MCP env configuration
- Query postgres data through tool
  - By default, make it readonly
  - Allow write ops by setting ENV `DANGEROUSLY_ALLOW_WRITE_OPS=true|1`
- Access tables as `resources`
- Use Zod for schema definitions
```

As you can see, this is quite a lightweight spec.Fee free to ad more details as needed.. Together with the README links, we’ll construct the final prompt

```markdown  theme={null}
Read the following and follow @spec.md to understand what we want. All necesary dependeies are installed
- @https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/refs/heads/main/README.md
- @https://raw.githubusercontent.com/porsager/postgres/refs/heads/master/README.md
```

With these three components in place (the specification, the MCP SDK documentation, and the Postgres library documentation), we can use Cursor to scaffold the server implementation. Cursor will help us stitch the pieces together, generating the code that connects the MCP SDK with Postgres.

After some back and forth prompting, we now have a first version of the MCP server going. To try it out, we can use the [MCP Inspector](https://modelcontextprotocol.io/legacy/tools/inspector)

```bash  theme={null}
npx @modelcontextprotocol/inspector bun run index.ts
```

## Testing the MCP Server

Once the initial implementation is complete, we can test it using the MCP Inspector. The inspector provides a way to see what the server exposes and to verify that the tools and resources behave as expected. We should confirm that queries can be executed and that schema information is returned correctly.

When everything looks good, we can connect the server to Cursor itself and test it in a real environment. At this point, Cursor will be able to use the Postgres MCP server as if it were a built-in capability, letting us query and inspect the database directly.

## Next Steps

Running the MCP server locally over stdio is a great starting point, but teams often require shared access to the same database through their MCP server. In these scenarios, deploying the MCP server as a centralized HTTP service becomes necessary.

A deployed MCP server offers several advantages over individual stdio instances:

* **Shared database access:** Multiple team members can query the same database instance through Cursor
* **Centralized configuration:** Schema updates and permission changes are managed in one location
* **Enhanced security:** Proper authentication, rate limiting, and access controls can be implemented
* **Observability:** Usage patterns and performance metrics can be monitored across the team

To achieve this, you would switch the transport method from stdio to HTTP.

While we won’t cover the whole setup, here’s a good starting prompt you can give to Cursor

```
Based on the existing MCP server, create a new file that implements the HTTP protocol.

Move shared logic to mcp-core, and name each transport implementation by name (mcp-server-stdio, mcp-server-http)

@https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/refs/heads/main/README.md 
```

The final results can be seen here: [pg-mcp-server](https://github.com/ericzakariasson/pg-mcp-server)

```mermaid  theme={null}
graph LR
    subgraph Team_Network_or_Cloud["VPC"]
      SVC["MCP Server (HTTP transport)"] <--> DB[(Postgres DB)]
    end

subgraph SVC_Internals["Internals"]
      direction TB
      T1["Auth and ACLs"]
      T2["Rate limiting and logging"]
      T3["Tool: query"]
      T4["Resource: table_schemas"]
      T5["Prompts"]
      T1 --> T2 --> T3
      T3 --> T4
      T3 --> T5
    end

SVC --> SVC_Internals
  
    class SVC svc
    class DB db
    class T1,T2,T3,T4,T5 internals
    class Dev1,Dev2,Dev3 dev
```

# Web Development
Source: https://docs.cursor.com/en/guides/tutorials/web-development

How to set up Cursor for web development

export const McpInstallButtonPrimary = ({server, showIcon = true, prompt = null}) => {
  const [showModal, setShowModal] = useState(false);
  const generateDeepLink = () => {
    if (!server || !server.name || !server.install) {
      return null;
    }
    try {
      if (typeof server.install === 'string') {
        return server.install;
      }
      if (server.install.url) {
        const config = {
          ...server.install
        };
        const jsonString = JSON.stringify(config);
        const utf8Bytes = new TextEncoder().encode(jsonString);
        const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
        const safeBase64Config = base64Config.replace(/\+/g, '%2B');
        return `cursor://anysphere.cursor-deeplink/mcp/install?name=${encodeURIComponent(server.name)}&config=${encodeURIComponent(safeBase64Config)}`;
      }
      if (server.install.command) {
        let config = {
          command: server.install.command,
          ...server.install.args && ({
            args: server.install.args
          }),
          ...server.install.env && ({
            env: server.install.env
          })
        };
        if (config.command && config.args) {
          const argsString = config.args.join(" ");
          config.command = `${config.command} ${argsString}`;
          delete config.args;
        }
        const jsonString = JSON.stringify(config);
        const utf8Bytes = new TextEncoder().encode(jsonString);
        const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
        const safeBase64Config = base64Config.replace(/\+/g, '%2B');
        return `cursor://anysphere.cursor-deeplink/mcp/install?name=${encodeURIComponent(server.name)}&config=${encodeURIComponent(safeBase64Config)}`;
      }
      return null;
    } catch (e) {
      console.error("Error generating deep link:", e);
      return null;
    }
  };
  const handleButtonClick = () => {
    setShowModal(true);
  };
  const handleClose = () => {
    setShowModal(false);
  };
  const deepLink = generateDeepLink();
  const isDocumentationOnly = typeof server?.install === 'string';
  const hasConfirmation = prompt || isDocumentationOnly;
  const InstallModal = ({isOpen, onClose, deepLink, server, children}) => {
    useEffect(() => {
      const handleKeyDown = event => {
        if (event.key === 'Escape') {
          onClose();
        }
      };
      if (isOpen) {
        document.addEventListener('keydown', handleKeyDown);
      }
      return () => {
        document.removeEventListener('keydown', handleKeyDown);
      };
    }, [isOpen, onClose]);
    if (!isOpen) return null;
    return <div className="fixed inset-0 bg-black bg-opacity-50 flex items-center justify-center z-50 transition-opacity duration-200" onClick={onClose}>
        <div className="bg-white dark:bg-neutral-900 rounded-lg p-6 max-w-md w-full mx-4 border border-neutral-200 dark:border-neutral-700 transition-all duration-200 transform" onClick={e => e.stopPropagation()}>
          <div className="mb-4">
            <h3 className="text-lg font-semibold text-black dark:text-white mb-2">
              Install {server?.name}
            </h3>
            <div className="text-neutral-600 dark:text-neutral-400">
              {children}
            </div>
          </div>
          <div className="flex gap-3 justify-end">
            <button onClick={onClose} className="px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 text-neutral-600 dark:text-neutral-400 hover:text-black dark:hover:text-white border border-neutral-200 dark:border-neutral-700 hover:bg-neutral-100 dark:hover:bg-neutral-800">
              Cancel
            </button>
            <a href={deepLink} onClick={onClose} target="_blank" className="px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 bg-black text-white hover:bg-neutral-800 dark:bg-white dark:text-black dark:hover:bg-neutral-200 inline-flex items-center justify-center no-underline">
              Continue
            </a>
          </div>
        </div>
      </div>;
  };
  return <>
      {hasConfirmation ? <button onClick={handleButtonClick} className="inline-flex justify-center items-center gap-2 px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 not-prose text-white bg-black hover:bg-neutral-800 dark:bg-white dark:text-black dark:hover:bg-neutral-200 border border-black dark:border-white">
          {showIcon && <Icon icon="plus" size={16} color="currentColor" />}
          Add to Cursor
        </button> : <a href={deepLink} className="inline-flex justify-center items-center gap-2 px-4 py-2 text-sm font-medium rounded-lg transition-colors duration-200 not-prose text-white bg-black hover:bg-neutral-800 dark:bg-white dark:text-black dark:hover:bg-neutral-200 border border-black dark:border-white">
          {showIcon && <Icon icon="plus" size={16} color="currentColor" />}
          Add to Cursor
        </a>}

{hasConfirmation && <InstallModal isOpen={showModal} onClose={handleClose} deepLink={deepLink} server={server}>
          {prompt}
        </InstallModal>}
    </>;
};

Web development involves fast iterations and tight feedback loops between Cursor and external tools like Figma or the browser. At Cursor, we've found workflows that tighten this loop. Clear task scoping, reusing components, and leveraging design systems help keep things fast and consistent.

This guide covers how to set up Cursor to support web development and tighten the feedback loop.

```mermaid  theme={null}
graph LR
  A[Project] --> B[Design]
  B --> C[Code]
  C -- Test --> D[Verify]
  D -- Feedback --> C
```

# Start orchestrating in Cursor

**Chat** is great for bootstrapping changes. Once the major pieces are in place, switching to **Inline Edit** and **Tab** helps maintain your flow state.

After setting up Cursor, you'll be able to orchestrate workflows across different tools. Below is a demonstration of what's possible: a snake game created by combining Linear, Figma, and browser tools. While real-world projects are typically more complex, this example showcases the potential of these integrated workflows.

# Connect to your project management tools

You can integrate Cursor into your existing project management software using different tooling. In this guide, we'll look at integrating Linear with their MCP server.

### Installation

Or manually add the Linear MCP server to `mcp.json`:

```jsx  theme={null}
{
  "mcpServers": {
    "Linear": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.linear.app/sse"
      ]
    }
  }
}
```

Then:

1. Make sure to enable Linear from MCP settings
2. Web browser will open and you will be prompted to authorize with Linear
   <Info>
     Due to the current state of MCP, the set up can require multiple attempts.
     If the integration is not working, try to "Reload" server from Cursor
     settings.
   </Info>

### Using Linear in Cursor

Linear MCP servers expose different tools that Cursor can use to read and manage issues. Go to MCP settings an locate the Linear server to see a list of all tools. To verify, try this prompt in Chat:

```jsx  theme={null}
list all issues related to this project
```

It should return a list of issues if the integration is set up properly.

# Bring in your Figma designs

Designs and mockups are core to web development. Using the official MCP server for Figma, you can directly access and work with design files in Cursor. To get started, follow the set up instructions at [Figma Dev Mode MCP Server](https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Dev-Mode-MCP-Server).

### Installation

You can click the button below to install the MCP server.

Or manually add to your `mcp.json`:

```jsx  theme={null}
{
  "mcpServers": {
    "Figma": {
      "url": "http://127.0.0.1:3845/sse"
    }
  }
}
```

### Usage

The server exposes multiple tools you can use in your prompts. E.g try to ask for the designs of the current selection in Figma. Read more in the [documentation](https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Dev-Mode-MCP-Server).

# Keep your code scaffolding consistent

You probably have existing code, a design system, or established conventions you want to reuse. When working with models, it's helpful to reference patterns already in your codebase, such as dropdown menus or other common components.

Working in a large web-based codebase ourselves, we've found that declarative code works especially well, particularly for React and JSX.

If you have a design system, you can help the agent discover it by providing a rule for it. Here's a `ui-components.mdc` file where we try to enforce reuse of components when possible:

```mdc components.mdc theme={null}
---
description: Implementing designs and building UI
---
- reuse existing UI components from `/src/components/ui`. these are the primitives we can build with
- create new components by orchestrating ui components if you can't find any existing that solves the problem
- ask the human how they want to proceed when there are missing components and designs
```

As your component library grows, add new rules accordingly. When the rules become too numerous, consider splitting them into more specific categories, such as "only apply when working with user inputs."

# Give Cursor access to browser

To extend Cursor's capabilities, you can set up the browser tools MCP server, which provides access to console logs and network requests. Once configured, you can verify your changes by monitoring console output and network activity. This set up helps ensure your implementation matches your intention. Follow the instructions here to set up the MCP server: [https://browsertools.agentdesk.ai/installation](https://browsertools.agentdesk.ai/installation)

<Note>
  We're working on making this easier to integrate natively in Cursor.
</Note>

# Takeaways

* Tight feedback loops are essential in web development. Use Cursor alongside tools like Figma, Linear, and the browser to move quickly and stay in flow.
* MCP servers let you integrate external systems directly into Cursor, reducing context switching and improving task execution.
* Reusing components and design systems helps the model produce cleaner, more consistent code and outputs.
* Clear, scoped tasks lead to better results. Be intentional with how you prompt and what you ask for.
* If you're not getting good outputs, try adjusting:
  * **Instructions:** Use rules, prompts, and give access to more context with e.g., MCP servers.
  * **Systems:** Patterns, abstractions, and clarity make it easier for the model to understand, help, and work more autonomously
* You can extend the model's context by including runtime info like console logs, network requests, and UI element data.
* Not everything needs to be automated. If your system becomes too complex, fall back to more surgical edits with Tab and Inline Edit.
* Cursor is most powerful when it's a co-pilot, not an autopilot. Use it to improve, not replace, your own decision-making.

# Working with Context
Source: https://docs.cursor.com/en/guides/working-with-context

How to work with context in Cursor

First, what is a context window? And how does it relate to effectively coding with Cursor?

To zoom out a bit, a large language model (LLM) is an artificial intelligence model trained to predict and generate text by learning patterns from massive datasets. It powers tools like Cursor by understanding your input and suggesting code or text based on what it’s seen before.

Tokens are the inputs and outputs of these models. They are chunks of text, often a fragment of a word, that an LLM processes one-by-one. Models don’t read entire sentences at once; they predict the next token based on the ones that came before.

To see how some text tokenizes, you can use a tokenizer like [this one](https://tiktokenizer.vercel.app/).

# What is context?

When we’re generating a code suggestion in Cursor, “context” refers to the information that is provided to the model (in the form of "input tokens") that the model then uses to predict the subsequent information (in the form of “output tokens”).

There are two types of context:

1. **Intent context** defines what the user wants to get out of the model. For example, a system prompt usually serves as high-level instructions for how the user wants the model to behave. Most of the "prompting" done in Cursor is intent context. “Turn that button from blue to green” is an example of stated intent; it is prescriptive.
2. **State context** describes the state of the current world. Providing Cursor with error messages, console logs, images, and chunks of code are examples of context related to state. It is descriptive, not prescriptive.

Together, these two types of context work in harmony by describing the current state and desired future state, enabling Cursor to make useful coding suggestions.

```mermaid  theme={null}
flowchart LR
    A["Intent (what you want)"] --> C[Model]
    B["State (what's true)"] --> C
    C -- Predict --> D["Action (what it does)"]
```

# Providing context in Cursor

The more relevant context you can provide a model, the more useful it will be. If insufficient context is provided in Cursor, the model will try to solve it without the relevant information. This typically results in:

1. Hallucinations where the model tries to pattern match (when there’s no pattern) causing unexpected results. This can happen frequently for models like `claude-3.5-sonnet` when it’s not given enough context.
2. The Agent trying to gather context by itself by searching the codebase, reading files, and calling tools. A strong thinking model (like `claude-3.7-sonnet`) can go quite far with this strategy, and providing the right initial context will determine the trajectory.

The good news is that Cursor is built with contextual awareness at its core and is designed to require minimal intervention from the user. Cursor automatically pulls in the parts of your codebase that the model estimates are relevant, such as the current file, semantically-similar patterns in other files, and other information from your session.

However, there’s a lot of context that can be pulled from, so manually specifying the context that you know is relevant to the task is a helpful way to steer the models in the right direction.

## @-symbol

The easiest way to provide explicit context is with the @-symbol. These are great when you know specifically what file, folder, website, or other piece of context you want to include. The more specific you can be, the better. Here’s a breakdown of how to get more surgical with context:

| Symbol    | Example              | Use case                                                                                | Drawback                                                                              |
| --------- | -------------------- | --------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------- |
| `@code`   | `@LRUCachedFunction` | You know which function, constant or symbol is relevant to the output you’re generating | Requires a lot of knowledge of codebase                                               |
| `@file`   | `cache.ts`           | You know which file should be read or edited, but not exactly where in the file         | Might include a lot of irrelevant context for the task at hand depending on file size |
| `@folder` | `utils/`             | Everything or majority of files in a folder is relevant                                 | Might include a lot of irrelevant context for the task at hand                        |

## Rules

You should think of rules as long-term memory that you want you or other members of your team to have access to. Capturing domain-specific context, including workflows, formatting and other conventions, is a great starting point for writing rules.

Rules can also be generated from existing conversations by using `/Generate Cursor Rules`. If you’ve had a long back and forth conversation with lots of prompting, there’s probably some useful directives or general rules that you might want to reuse later.

## MCP

[Model Context Protocol](https://modelcontextprotocol.io/introduction) is an extensibility layer where you can give Cursor capabilities to perform actions and pull in external context.

Depending on your development setup, you might want to leverage different types of servers, but two categories that we’ve seen be particularly useful are:

* **Internal documentation**: e.g., Notion, Confluence, Google Docs
* **Project management**: e.g., Linear, Jira

If you have existing tooling for accessing context and performing actions through an API, you can build an MCP server for it. Here’s a short guide on how to build [MCP servers](https://modelcontextprotocol.io/tutorials/building-mcp-with-llms).

## Self-gathering context

A powerful pattern many users are adopting is letting the Agent write short-lived tools that it can then run to gather more context. This is especially effective in human-in-the-loop workflows where you review the code before it’s executed.

For example, adding debugging statements to your code, running it, and letting the model inspect the output gives it access to dynamic context it couldn’t infer statically.

In Python, you can do this by prompting the Agent to:

1. Add print("debugging: ...") statements in relevant parts of the code
2. Run the code or tests using the terminal

The Agent will read the terminal output and decide what to do next. The core idea is to give the Agent access to the actual runtime behavior, not just the static code.

# Takeaways

* Context is the foundation of effective AI coding, consisting of intent (what you want) and state (what exists). Providing both helps Cursor make accurate predictions.
* Use surgical context with @-symbols (@code, @file, @folder) to guide Cursor precisely, rather than relying solely on automatic context gathering.
* Capture repeatable knowledge in rules for team-wide reuse, and extend Cursor's capabilities with Model Context Protocol to connect external systems.
* Insufficient context leads to hallucinations or inefficiency, while too much irrelevant context dilutes the signal. Strike the right balance for optimal results.

# Inline Edit
Source: https://docs.cursor.com/en/inline-edit/overview

Edit and ask questions with Inline Edit (Cmd/Ctrl+K) in Cursor

Inline Edit lets you edit code or ask questions directly in your editor with <Kbd>Cmd+K</Kbd>, which opens an input field where your selected code and instructions create your request.

## Modes

### Edit Selection

With code selected, <Kbd>Cmd+K</Kbd> edits that specific code based on your instructions.

Without selection, Cursor generates new code at your cursor position. The AI includes relevant surrounding code for context. For example, triggering on a function name includes the entire function.

### Quick Question

Press <Kbd>Opt+Return</Kbd> in the inline editor to ask questions about selected code.

After getting an answer, type "do it" or similar wording to convert the suggestion into code. This lets you explore ideas before implementing.

### Full File Edits

For file-wide changes, use <Kbd>Cmd+Shift+Return</Kbd>. This mode enables comprehensive changes while maintaining control.

### Send to Chat

For multi-file edits or advanced features, use <Kbd>Cmd+L</Kbd> to send selected code to [Chat](/en/agent/modes#agent). This provides multi-file editing, detailed explanations, and advanced AI capabilities.

## Follow-up instructions

After each edit, refine results by adding instructions and pressing <Kbd>Return</Kbd>. The AI updates changes based on your feedback.

## Default context

Inline Edit includes default context to improve code generation beyond any [@ symbols](/en/context/@-symbols/@-files) you add.

This includes related files, recently viewed code, and relevant information. Cursor prioritizes the most relevant context for better results.

# Terminal
Source: https://docs.cursor.com/en/inline-edit/terminal

Generate terminal commands with Cmd/Ctrl+K

In Cursor terminal, press <Kbd>Cmd+K</Kbd> to open a prompt bar at the bottom.
Describe your desired action and Inline Edit generates a command.

Inline Edit in terminal uses your recent terminal history, instructions, and prompt content as context.

# Git
Source: https://docs.cursor.com/en/integrations/git

AI-powered Git features including commit message generation and merge conflict resolution

Cursor provides AI-powered Git features to streamline your workflow, including automatic commit message generation and intelligent merge conflict resolution.

## AI Commit Message

Cursor generates commit messages from staged changes.

1. Stage files to commit
2. Open the Git tab in the sidebar
3. Click the sparkle (✨) icon next to the commit message input

Generated messages use staged changes and repository git history. If you use conventions like [Conventional Commits](https://www.conventionalcommits.org/), messages follow the same pattern.

### Add shortcut

To bind to a keyboard shortcut:

1. Go to Keyboard Shortcuts (<Kbd>Cmd+R Cmd+S</Kbd> or <Kbd>Cmd+Shift+P</Kbd> and search "Open Keyboard Shortcuts (JSON)")
2. Add this binding for <Kbd>Cmd+M</Kbd>:
   ```json  theme={null}
   {
     "key": "cmd+m",
     "command": "cursor.generateGitCommitMessage"
   }
   ```
3. Save

<Info>
  You cannot customize commit message generation. Cursor adapts to your existing commit style.
</Info>

## AI Resolve Conflicts

When merge conflicts occur, Cursor Agent can help resolve them by understanding both sides of the conflict and proposing a resolution.

### How to use

1. When a merge conflict occurs, you'll see the conflict markers in your file
2. Click the **Resolve in Chat** button that appears in the merge conflict UI
3. Agent will analyze both versions and suggest a resolution
4. Review the proposed changes and apply them

# GitHub
Source: https://docs.cursor.com/en/integrations/github

Official Cursor GitHub app for background agents

[Background Agents](/en/background-agent) and [Bugbot](/en/bugbot) require the Cursor GitHub app to clone repositories and push changes.

## Installation

1. Go to [Integrations in Dashboard](https://cursor.com/dashboard?tab=integrations)
2. Click **Connect** next to GitHub
3. Choose repository either **All repositories** or **Selected repositories**

To disconnect your GitHub account, return to the integrations dashboard and click **Disconnect Account**.

## Using Agent in GitHub

The GitHub integration enables background agent workflows directly from pull requests and issues. You can trigger an agent to read context, implement fixes, and push commits by commenting `@cursor [prompt]` on any PR or issue.

If you have [Bugbot](/en/bugbot) enabled, you can comment `@cursor fix` to read the suggested fix from Bugbot to trigger a background agent to address the issue.

## Permissions

The GitHub app requires specific permissions to work with background agents:

<div className="full-width-table">
  | Permission                | Purpose                                          |
  | ------------------------- | ------------------------------------------------ |
  | **Repository access**     | Clone your code and create working branches      |
  | **Pull requests**         | Create PRs with agent changes for your review    |
  | **Issues**                | Track bugs and tasks that agents discover or fix |
  | **Checks and statuses**   | Report on code quality and test results          |
  | **Actions and workflows** | Monitor CI/CD pipelines and deployment status    |
</div>

All permissions follow the principle of least privilege needed for background agent functionality.

## IP Allow List Configuration

If your organization uses GitHub's IP allow list feature to restrict access to your repositories, you'll need to contact support first to enable IP allowlist functionality for your team.

### Contact Support

Before configuring IP allowlists, contact [hi@cursor.com](mailto:hi@cursor.com) to enable this feature for your team. This is required for both configuration methods below.

### Enable IP allow list configuration for installed GitHub Apps (Recommended)

The Cursor GitHub app has the IP list already pre-configured. You can enable the allowlist for installed apps to automatically inherit this list. This is the **recommended approach**, as it allows us to update the list and your organization receives updates automatically.

To enable this:

1. Go to your organization's Security settings
2. Navigate to IP allow list settings
3. Check **"Allow access by GitHub Apps"**

For detailed instructions, see [GitHub's documentation](https://docs.github.com/en/enterprise-cloud@latest/organizations/keeping-your-organization-secure/managing-security-settings-for-your-organization/managing-allowed-ip-addresses-for-your-organization#allowing-access-by-github-apps).

### Add IPs directly to your allowlist

If your organization uses IdP-defined allowlists in GitHub or otherwise cannot use the pre-configured allowlist, you can manually add the IP addresses:

```
184.73.225.134
3.209.66.12
52.44.113.131
```

<Note>
  The list of IP addresses may infrequently change. Teams using IP allow lists will be given advanced notice before IP addresses are added or removed.
</Note>

## Troubleshooting

<AccordionGroup>
  <Accordion title="Agent can't access repository">
    * Install the GitHub app with repository access
    * Check repository permissions for private repos
    * Verify your GitHub account permissions
  </Accordion>

<Accordion title="Permission denied for pull requests">
    * Grant the app write access to pull requests
    * Check branch protection rules
    * Reinstall if the app installation expired
  </Accordion>

<Accordion title="App not visible in GitHub settings">
    * Check if installed at organization level
    * Reinstall from [github.com/apps/cursor](https://github.com/apps/cursor)
    * Contact support if installation is corrupted
  </Accordion>
</AccordionGroup>

# Linear
Source: https://docs.cursor.com/en/integrations/linear

Work with Background Agents from Linear

Use [Background Agents](/en/background-agent) directly from Linear by delegating issues to Cursor or mentioning `@Cursor` in comments.

## Get started

### Installation

<Note>
  You must be a Cursor admin to connect the Linear integration. Other team settings are available to non-admin members.
</Note>

1. Go to [Cursor integrations](https://www.cursor.com/en/dashboard?tab=integrations)
2. Click *Connect* next to Linear
3. Connect your Linear workspace and select team
4. Click *Authorize*
5. Complete any remaining Background Agent setup in Cursor:
   * Connect GitHub and select default repository
   * Enable usage-based pricing
   * Confirm privacy settings

### Account linking

First use prompts account linking between Cursor and Linear. GitHub connection required for PR creation.

## How to use

Delegate issues to Cursor or mention `@Cursor` in comments. Cursor analyzes issues and filters out non-development work automatically.

### Delegating issues

1. Open Linear issue
2. Click assignee field
3. Select "Cursor"

### Mentioning Cursor

Mention `@Cursor` in a comment to assign a new agent or provide additional instructions, for example: `@Cursor fix the authentication bug described above`.

## Workflow

Background Agents show real-time status in Linear and create PRs automatically when complete. Track progress in [Cursor dashboard](https://www.cursor.com/dashboard?tab=background-agents).

### Follow-up instructions

You can respond in the agent session and it'll get sent as a follow-up to the agent. Simply mention `@Cursor` in a Linear comment to provide additional guidance to a running Background Agent.

## Configuration

Configure Background Agent settings from [Dashboard → Background Agents](https://www.cursor.com/dashboard?tab=background-agents).

### Configuration options

You can configure Background Agent behavior using several methods:

**Issue description or comments**: Use `[key=value]` syntax, for example:

* `@cursor please fix [repo=anysphere/everysphere]`
* `@cursor implement feature [model=claude-3.5-sonnet] [branch=feature-branch]`

**Issue labels**: Use parent-child label structure where the parent label is the configuration key and the child label is the value.

**Project labels**: Same parent-child structure as issue labels, applied at the project level.

Supported configuration keys:

* `repo`: Specify target repository (e.g., `owner/repository`)
* `branch`: Specify base branch for PR creation
* `model`: Specify AI model to use

### Repository selection

Cursor determines which repository to work on using this priority order:

1. **Issue description/comments**: `[repo=owner/repository]` syntax in issue text or comments
2. **Issue labels**: Repository labels attached to the specific Linear issue
3. **Project labels**: Repository labels attached to the Linear project
4. **Default repository**: Repository specified in Cursor dashboard settings

#### Setting up repository labels

To create repository labels in Linear:

1. Go to **Settings** in your Linear workspace
2. Click **Labels**
3. Click **New group**
4. Name the group "repo" (case insensitive - must be exactly "repo", not "Repository" or other variations)
5. Within that group, create labels for each repository using the format `owner/repo`

These labels can then be assigned to issues or projects to specify which repository the Background Agent should work on.

{/* ### Getting help

Check [agent activity](https://www.cursor.com/dashboard?tab=background-agents) and include request IDs when contacting support.

## Feedback

Share feedback through Linear comments or your Cursor dashboard support channels. */}

# Slack
Source: https://docs.cursor.com/en/integrations/slack

Work with Background Agents from Slack

export const SlackThread = ({messages = []}) => {
  const MessageWithMentions = ({text}) => {
    const parts = text.split(/(@\w+)/g);
    return <>
        {parts.map((part, index) => {
      if (part.startsWith('@')) {
        return <span key={index} className="text-[#1264A3] bg-[#1264A3]/10 dark:bg-[#1264A3]/25 px-0.5 py-0.5 rounded hover:bg-[#1264A3]/20 cursor-pointer transition-colors">
                {part}
              </span>;
      }
      return <span key={index}>{part}</span>;
    })}
      </>;
  };
  return <div className="border border-neutral-200 dark:border-neutral-700 rounded-lg bg-neutral-50 dark:bg-neutral-900/50 py-4 overflow-hidden">
      {messages.map((msg, index) => <div key={index} className={`group hover:bg-[#f0f0f0] dark:hover:bg-[#333] px-6 py-2 -mx-2 -my-1 transition-colors`}>
          <div className="flex items-start gap-3">
            <div className="w-9 h-9 rounded-md bg-neutral-300 dark:bg-neutral-800 flex items-center justify-center text-white text-sm font-semibold flex-shrink-0">
              {msg.name ? msg.name.charAt(0).toUpperCase() : 'U'}
            </div>

<div className="flex-1 min-w-0">
              <div className="flex items-baseline gap-2">
                <span className="font-semibold text-neutral-900 dark:text-neutral-100 text-sm">
                  {msg.name || 'User'}
                </span>
                <span className="text-xs text-neutral-500 dark:text-neutral-400">
                  {msg.timestamp || ''}
                </span>
              </div>
              <div className="text-neutral-900 dark:text-neutral-100 text-[15px] leading-relaxed">
                <MessageWithMentions text={msg.message} />
              </div>

{msg.reactions && msg.reactions.length > 0 && <div className="flex gap-1 mt-1">
                  {msg.reactions.map((reaction, rIndex) => <div key={rIndex} className="inline-flex items-center gap-0.5 px-1.5 py-0.5 bg-white dark:bg-neutral-800 border border-neutral-200 dark:border-neutral-700 rounded text-xs hover:bg-neutral-100 dark:hover:bg-neutral-700 transition-colors cursor-pointer">
                      <span>{reaction.emoji}</span>
                      <span className="text-neutral-600 dark:text-neutral-400">{reaction.count}</span>
                    </div>)}
                </div>}
            </div>
          </div>
        </div>)}
    </div>;
};

export const SlackInlineMessage = ({message}) => {
  const MessageWithMentions = ({text}) => {
    const parts = text.split(/(@\w+)/g);
    return <>
        {parts.map((part, index) => {
      if (part.startsWith('@')) {
        return <span key={index} className="text-[#1264A3] hover:bg-[#1264A3]/10 dark:hover:bg-[#1264A3]/25 px-0.5 rounded">
                {part}
              </span>;
      }
      return <span key={index}>{part}</span>;
    })}
      </>;
  };
  return <span className="inline rounded p-0.5 bg-neutral-50 dark:bg-neutral-800/30">
      <MessageWithMentions text={message} />
    </span>;
};

export const SlackUserMessage = ({message, reactions = [], replies = null}) => {
  const MessageWithMentions = ({text}) => {
    const parts = text.split(/(@\w+)/g);
    return <>
        {parts.map((part, index) => {
      if (part.startsWith('@')) {
        return <span key={index} className="text-[#1264A3] bg-[#1264A3]/10 dark:bg-[#1264A3]/25 px-0.5 py-0.5 rounded hover:bg-[#1264A3]/20 cursor-pointer transition-colors">
                {part}
              </span>;
      }
      return <span key={index}>{part}</span>;
    })}
      </>;
  };
  return <div className="border border-neutral-200 dark:border-neutral-700 rounded-lg hover:bg-neutral-50 dark:hover:bg-neutral-800/50 transition-colors px-5 py-3 group">
      <div className="text-neutral-900 dark:text-neutral-100 text-[15px] leading-relaxed">
        <MessageWithMentions text={message} />
      </div>

{reactions.length > 0 && <div className="flex gap-1 mt-1">
          {reactions.map((reaction, index) => <div key={index} className="inline-flex items-center gap-0.5 px-1.5 py-0.5 bg-neutral-100 dark:bg-neutral-800 rounded text-xs hover:bg-neutral-200 dark:hover:bg-neutral-700 transition-colors cursor-pointer">
              <span>{reaction.emoji}</span>
              <span className="text-neutral-600 dark:text-neutral-400">{reaction.count}</span>
            </div>)}
        </div>}

{replies && <div className="flex items-center gap-1.5 mt-2 text-[#1264A3] hover:underline cursor-pointer">
          <svg className="w-4 h-4" fill="currentColor" viewBox="0 0 20 20">
            <path d="M7.707 10.293a1 1 0 10-1.414 1.414l3 3a1 1 0 001.414 0l3-3a1 1 0 00-1.414-1.414L11 11.586V6h5a2 2 0 012 2v7a2 2 0 01-2 2H4a2 2 0 01-2-2V8a2 2 0 012-2h5v5.586l-1.293-1.293z" />
          </svg>
          <span className="text-sm font-medium">{replies.count} {replies.count === 1 ? 'reply' : 'replies'}</span>
          {replies.lastReplyTime && <span className="text-xs text-neutral-500 dark:text-neutral-400">{replies.lastReplyTime}</span>}
        </div>}
    </div>;
};

With Cursor's integration for Slack, you can use [Background Agents](/en/background-agent) to work on your tasks directly from Slack by mentioning <SlackInlineMessage message="@Cursor" /> with a prompt.

## Get started

### Installation

1. Go to [Cursor integrations](https://www.cursor.com/en/dashboard?tab=integrations)

2. Click *Connect* next to Slack or go to [installation page](https://cursor.com/api/install-slack-app) from here

3. You'll be prompted to install the Cursor app for Slack in your workspace.

4. After installing in Slack, you'll be redirected back to Cursor to finalize setup

1. Connect GitHub (if not already connected) and pick a default repository
   2. Enable usage-based pricing
   3. Confirm privacy settings

5. Start using Background Agents in Slack by mentioning <SlackInlineMessage message="@Cursor" />

## How to use

Mention <SlackInlineMessage message="@Cursor" /> and give your prompt. This handles most use cases, but you can also use commands below to customize your agent.

For example, mention <SlackInlineMessage message="@Cursor fix the login bug" /> directly in conversation, or use specific commands like <SlackInlineMessage message="@Cursor [repo=torvalds/linux] fix bug" /> to target a particular repository.

### Commands

Run <SlackInlineMessage message="@Cursor help" /> for an up-to-date command list.

<div className="full-width-table">
  | Command                                                     | Description                                                                           |
  | :---------------------------------------------------------- | :------------------------------------------------------------------------------------ |
  | <SlackInlineMessage message="@Cursor [prompt]" />           | Start a Background Agent. In threads with existing agents, adds followup instructions |
  | <SlackInlineMessage message="@Cursor settings" />           | Configure defaults and channel's default repository                                   |
  | <SlackInlineMessage message="@Cursor [options] [prompt]" /> | Use advanced options: `branch`, `model`, `repo`                                       |
  | <SlackInlineMessage message="@Cursor agent [prompt]" />     | Force create a new agent in a thread                                                  |
  | <SlackInlineMessage message="@Cursor list my agents" />     | Show your running agents                                                              |
</div>

#### Options

Customize Background Agent behavior with these options:

##### Syntax Formats

Use options in several ways:

1. **Bracket format**

2. **Inline format**
   <SlackInlineMessage message="@Cursor branch=dev model=o3 repo=owner/repo autopr=false Fix the login bug" />

##### Option precedence

When combining options:

* **Explicit values** override defaults
* **Later values** override earlier ones if duplicated
* **Inline options** take precedence over settings modal defaults

The bot parses options from anywhere in the message, allowing natural command writing.

#### Using thread context

Background Agents understand and use context from existing thread discussions. Useful when your team discusses an issue and you want the agent to implement the solution based on that conversation.

<Note>
  Background Agents read the entire thread for context when invoked,
  understanding and implementing solutions based on the team's discussion.
</Note>

#### When to use force commands

**When do I need <SlackInlineMessage message="@Cursor agent" />?**

In threads with existing agents, <SlackInlineMessage message="@Cursor [prompt]" /> adds followup instructions (only works if you own the agent). Use <SlackInlineMessage message="@Cursor agent [prompt]" /> to launch a separate agent.

**When do I need `Add follow-up` (from context menu)?**

Use the context menu (⋯) on an agent's response for followup instructions. Useful when multiple agents exist in a thread and you need to specify which one to follow up on.

### Status updates & handoff

When Background Agent runs, you first get an option to *Open in Cursor*.

When Background Agent completes, you get a notification in Slack and an option to view the created PR in GitHub.

### Managing agents

To see all running agents, run <SlackInlineMessage message="@Cursor list my agents" />.

Manage Background Agents using the context menu by clicking the three dots (⋯) on any agent message.

Available options:

* **Add follow-up**: Add instructions to an existing agent
* **Delete**: Stop and archive the Background Agent
* **View request ID**: View unique request ID for troubleshooting (include when contacting support)
* **Give feedback**: Provide feedback about agent performance

## Configuration

Manage default settings and privacy options from [Dashboard → Background Agents](https://www.cursor.com/dashboard?tab=background-agents).

### Settings

#### Default Model

Used when no model is explicitly specified with <SlackInlineMessage message="@Cursor [model=...]" />. See [settings](https://www.cursor.com/dashboard?tab=background-agents) for available options.

#### Default Repository

Used when no repository is specified. Use these formats:

* `https://github.com/org/repository`
* `org/repository`

<Note>
  If you reference a non-existent repository, it appears as if you don't have
  access. This shows in the error message when Background Agent fails to start.
</Note>

#### Base Branch

Starting branch for Background Agent. Leave blank to use the repository's default branch (often `main`)

### Channel Settings

Configure default settings at the channel level using <SlackInlineMessage message="@Cursor settings" />. These settings are per team and override your personal defaults for that channel.

Particularly useful when:

* Different channels work on different repositories
* Teams want consistent settings across all members
* You want to avoid specifying the repository in every command

To configure channel settings:

1. Run <SlackInlineMessage message="@Cursor settings" /> in the desired channel
2. Set the default repository for that channel
3. All team members using Background Agents in that channel use these defaults

<Note>
  Channel settings take precedence over personal defaults but can be overridden
  by explicit options like{" "}

### Privacy

Background Agents support Privacy Mode.

Read more about [Privacy Mode](https://www.cursor.com/privacy-overview) or manage your [privacy settings](https://www.cursor.com/dashboard?tab=background-agents).

<Warning>
  Privacy Mode (Legacy) is not supported. Background Agents require temporary
  code storage while running.
</Warning>

#### Display Agent Summary

Display agent summaries and diff images. May contain file paths or code snippets. Can be turned On/Off.

#### Display Agent Summary in External Channels

For Slack Connect with other workspaces or channels with external members like Guests, choose to display agent summaries in external channels.

## Permissions

Cursor requests these Slack permissions for Background Agents to work within your workspace:

# Models
Source: https://docs.cursor.com/en/models

Available models in Cursor

export const ModelsTable = ({isPricing}) => {
  const MODEL_LIST = [{
    "id": "claude-4-sonnet",
    "name": "Claude 4 Sonnet",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/sonnet",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.3,
    "tokenOutput": 15,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "subRows": [{
      "id": "claude-4-sonnet-thinking",
      "name": "Thinking",
      "requests": 2
    }]
  }, {
    "id": "claude-4-sonnet-1m",
    "name": "Claude 4 Sonnet 1M",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/sonnet",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.3,
    "tokenOutput": 15,
    "contextWindow": "-",
    "maxContextWindow": "1M",
    "isMax": "only",
    "thinking": true,
    "badges": [],
    "notes": ["This model can be very expensive due to the large context window", "The cost is 2x when the input exceeds 200k tokens"],
    "subRows": [{
      "id": "claude-4-sonnet-1m-thinking",
      "name": "Thinking",
      "requests": 2
    }]
  }, {
    "id": "claude-4-opus",
    "name": "Claude 4 Opus",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/sonnet",
    "isAgent": true,
    "requests": 0.75,
    "tokenInput": 15,
    "tokenInputCached": 1.5,
    "tokenOutput": 75,
    "contextWindow": "-",
    "maxContextWindow": "200k",
    "isMax": "only",
    "thinking": true,
    "hidden": true,
    "badges": [],
    "notes": []
  }, {
    "id": "claude-4.1-opus",
    "name": "Claude 4.1 Opus",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/opus",
    "isAgent": true,
    "requests": 0.75,
    "tokenInput": 15,
    "tokenInputCached": 1.5,
    "tokenOutput": 75,
    "contextWindow": "-",
    "maxContextWindow": "200k",
    "isMax": "only",
    "thinking": true,
    "badges": [],
    "notes": []
  }, {
    "id": "claude-3.7-sonnet",
    "name": "Claude 3.7 Sonnet",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/sonnet",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.3,
    "tokenOutput": 15,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "trait": "Powerful but eager to make changes",
    "hidden": true,
    "subRows": [{
      "id": "claude-3.7-sonnet-thinking",
      "name": "Thinking",
      "requests": 2,
      "notes": ["More requests due to token intensive"]
    }]
  }, {
    "id": "claude-3.5-sonnet",
    "name": "Claude 3.5 Sonnet",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/sonnet",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.3,
    "tokenOutput": 15,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "trait": "Great all rounder for most tasks",
    "hidden": true
  }, {
    "id": "claude-3.5-haiku",
    "name": "Claude 3.5 Haiku",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/claude/haiku",
    "isAgent": false,
    "requests": 0.3333333333333333,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "claude-3-opus",
    "name": "Claude 3 Opus",
    "provider": "Anthropic",
    "link": "https://www.anthropic.com/news/claude-3-family",
    "isAgent": false,
    "requests": 2.5,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "cursor-small",
    "name": "Cursor Small",
    "provider": "Cursor",
    "link": null,
    "isAgent": false,
    "requests": 0,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "deepseek-v3",
    "name": "Deepseek V3",
    "provider": "DeepSeek",
    "link": "https://www.deepseek.com/",
    "isAgent": true,
    "requests": 0.3333333333333333,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "deepseek-v3-1",
    "name": "Deepseek V3.1",
    "provider": "DeepSeek",
    "link": "https://www.deepseek.com/",
    "isAgent": true,
    "requests": 0.3333333333333333,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "deepseek-r1",
    "name": "Deepseek R1",
    "provider": "DeepSeek",
    "link": "https://www.deepseek.com/",
    "isAgent": false,
    "requests": 1,
    "tokenInput": 1,
    "tokenInputCached": 2,
    "tokenOutput": 1,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "deepseek-r1-0528",
    "name": "Deepseek R1 (05/28)",
    "provider": "DeepSeek",
    "link": "https://www.deepseek.com/",
    "isAgent": false,
    "requests": 1,
    "tokenInput": 1,
    "tokenInputCached": 2,
    "tokenOutput": 1,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "gemini-2.5-pro-exp",
    "name": "Gemini 2.5 Pro",
    "provider": "Google",
    "link": "https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 1.25,
    "tokenInputCached": 0.31,
    "tokenOutput": 10,
    "docs": "https://ai.google.dev/gemini-api/docs/pricing",
    "contextWindow": "200k",
    "maxContextWindow": "1M",
    "thinking": true,
    "isMax": true,
    "badges": [],
    "notes": [],
    "trait": "Careful and precise",
    "subRows": [{
      "id": "gemini-2.5-pro-exp-long",
      "name": "Long Context (>200k)",
      "tokenInput": 2.5,
      "tokenInputCached": 0.625,
      "tokenOutput": 15,
      "isMax": true
    }]
  }, {
    "id": "gemini-2.5-flash-preview-5-20",
    "name": "Gemini 2.5 Flash",
    "provider": "Google",
    "link": "https://developers.googleblog.com/en/start-building-with-gemini-25-flash/",
    "isAgent": true,
    "thinking": true,
    "requests": 0,
    "tokenInput": 0.15,
    "tokenInputCached": 0.0375,
    "tokenOutput": 3.5,
    "docs": "https://ai.google.dev/gemini-api/docs/pricing",
    "contextWindow": "-",
    "maxContextWindow": "1M",
    "isMax": "only",
    "badges": [],
    "notes": []
  }, {
    "id": "gemini-2.0-pro-exp",
    "name": "Gemini 2.0 Pro (exp)",
    "provider": "Google",
    "link": "https://blog.google/technology/google-deepmind/gemini-model-updates-february-2025/",
    "isAgent": false,
    "requests": 1,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "thinking": true,
    "isMax": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "gpt-4o",
    "name": "GPT-4o",
    "provider": "OpenAI",
    "link": "https://openai.com/index/hello-gpt-4o/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 2.5,
    "tokenInputCached": 1.25,
    "tokenOutput": 10,
    "docs": "https://platform.openai.com/docs/models/gpt-4o",
    "contextWindow": "128k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "gpt-4o-mini",
    "name": "GPT-4o mini",
    "provider": "OpenAI",
    "link": "https://openai.com/gpt-4o-mini",
    "isAgent": false,
    "requests": 0,
    "tokenInput": 0.15,
    "tokenInputCached": 0.075,
    "tokenOutput": 0.6,
    "docs": "https://platform.openai.com/docs/models/gpt-4o-mini",
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": false,
    "badges": [],
    "notes": ["500 requests/day with free plan"],
    "hidden": true
  }, {
    "id": "gpt-4.5-preview",
    "name": "GPT 4.5 Preview",
    "provider": "OpenAI",
    "link": "https://openai.com/index/introducing-gpt-4-5/",
    "isAgent": false,
    "requests": 50,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "gpt-4.1",
    "name": "GPT 4.1",
    "provider": "OpenAI",
    "link": "https://openai.com/index/gpt-4-1/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 2,
    "tokenInputCached": 0.5,
    "tokenOutput": 8,
    "contextWindow": "200k",
    "maxContextWindow": "1M",
    "thinking": false,
    "isMax": true,
    "badges": [],
    "notes": []
  }, {
    "id": "gpt-5",
    "name": "GPT-5",
    "provider": "OpenAI",
    "link": "https://openai.com/index/gpt-5/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.75,
    "tokenOutput": 15,
    "docs": "https://platform.openai.com/docs/models/gpt-5",
    "contextWindow": "272k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": ["Agentic and reasoning capabilities", "Available reasoning effort variant is gpt-5-high"],
    "trait": "Advanced reasoning and agent capabilities"
  }, {
    "id": "gpt-5-nano",
    "name": "GPT-5 Nano",
    "provider": "OpenAI",
    "link": "https://openai.com/index/gpt-5/",
    "isAgent": true,
    "requests": 0.25,
    "tokenInput": 3,
    "tokenInputCached": 0.75,
    "tokenOutput": 15,
    "docs": "https://platform.openai.com/docs/models/gpt-5",
    "contextWindow": "272k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "hidden": true,
    "badges": [],
    "notes": []
  }, {
    "id": "gpt-5-mini",
    "name": "GPT-5 Mini",
    "provider": "OpenAI",
    "link": "https://openai.com/index/gpt-5/",
    "isAgent": true,
    "requests": 0.5,
    "tokenInput": 3,
    "tokenInputCached": 0.75,
    "tokenOutput": 15,
    "docs": "https://platform.openai.com/docs/models/gpt-5",
    "contextWindow": "272k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "hidden": true,
    "badges": [],
    "notes": []
  }, {
    "id": "gpt-5-fast",
    "name": "GPT-5 Fast",
    "provider": "OpenAI",
    "link": "https://openai.com/index/gpt-5/",
    "isAgent": true,
    "requests": 0.75,
    "tokenInput": 3,
    "tokenInputCached": 0.75,
    "tokenOutput": 15,
    "docs": "https://platform.openai.com/docs/models/gpt-5",
    "contextWindow": "272k",
    "maxContextWindow": "-",
    "isMax": false,
    "thinking": true,
    "badges": [],
    "notes": ["Faster speed but 2x price", "Available reasoning effort variants are gpt-5-high-fast, gpt-5-low-fast"]
  }, {
    "id": "o1",
    "name": "o1",
    "provider": "OpenAI",
    "link": "https://openai.com/index/learning-to-reason-with-llms/",
    "isAgent": false,
    "requests": 10,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "o1-mini",
    "name": "o1 Mini",
    "provider": "OpenAI",
    "link": "https://openai.com/index/openai-o1-mini-advancing-cost-efficient-reasoning/",
    "isAgent": false,
    "requests": 2.5,
    "contextWindow": "128k",
    "maxContextWindow": "-",
    "thinking": true,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "o3",
    "name": "o3",
    "provider": "OpenAI",
    "link": "https://openai.com/index/introducing-o3-and-o4-mini/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 2,
    "tokenInputCached": 0.5,
    "tokenOutput": 8,
    "docs": "https://platform.openai.com/docs/models/o3",
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "thinking": true,
    "isMax": true,
    "badges": [],
    "hidden": true,
    "notes": ["High reasoning effort"],
    "description": "For the most complex tasks"
  }, {
    "id": "o3-mini",
    "name": "o3-mini",
    "provider": "OpenAI",
    "link": "https://openai.com/index/openai-o3-mini/",
    "isAgent": true,
    "requests": 0.25,
    "tokenInput": 1.1,
    "tokenInputCached": 0.55,
    "tokenOutput": 4.4,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "thinking": true,
    "badges": [],
    "notes": ["High reasoning effort"],
    "hidden": true
  }, {
    "id": "o4-mini",
    "name": "o4-mini",
    "provider": "OpenAI",
    "link": "https://openai.com/index/introducing-o3-and-o4-mini/",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 1.1,
    "tokenInputCached": 0.275,
    "tokenOutput": 4.4,
    "contextWindow": "200k",
    "maxContextWindow": "-",
    "isMax": true,
    "hidden": true,
    "thinking": true,
    "badges": [],
    "notes": ["High reasoning effort"]
  }, {
    "id": "grok-2",
    "name": "Grok 2",
    "provider": "xAI",
    "link": "https://x.ai/blog/grok-1212",
    "isAgent": false,
    "requests": 1,
    "contextWindow": "60k",
    "maxContextWindow": "-",
    "thinking": false,
    "badges": [],
    "notes": [],
    "hidden": true
  }, {
    "id": "grok-3-beta",
    "name": "Grok 3 Beta",
    "provider": "xAI",
    "link": "https://x.ai/news/grok-3",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenOutput": 15,
    "contextWindow": "132k",
    "maxContextWindow": "-",
    "thinking": true,
    "isMax": true,
    "badges": [],
    "hidden": true,
    "notes": [],
    "docs": "https://docs.x.ai/docs/models#models-and-pricing"
  }, {
    "id": "grok-3-mini",
    "name": "Grok 3 Mini",
    "provider": "xAI",
    "link": "https://x.ai/news/grok-3",
    "isAgent": true,
    "requests": 0,
    "contextWindow": "132k",
    "maxContextWindow": "-",
    "thinking": false,
    "isMax": true,
    "tokenInput": 0.3,
    "tokenInputCached": 0.3,
    "tokenOutput": 1,
    "hidden": true,
    "badges": [],
    "notes": [],
    "docs": "https://docs.x.ai/docs/models#models-and-pricing"
  }, {
    "id": "grok-4",
    "name": "Grok 4",
    "provider": "xAI",
    "link": "https://docs.x.ai/docs/models/grok-4",
    "isAgent": true,
    "requests": 1,
    "tokenInput": 3,
    "tokenInputCached": 0.75,
    "tokenOutput": 15,
    "contextWindow": "256k",
    "maxContextWindow": "-",
    "thinking": true,
    "isMax": true,
    "hidden": true,
    "badges": [],
    "notes": [],
    "docs": "https://docs.x.ai/docs/models/grok-4"
  }, {
    "id": "grok-code-fast-1",
    "name": "Grok Code",
    "provider": "xAI",
    "isAgent": true,
    "link": "https://docs.x.ai/docs/models#models-and-pricing",
    "requests": 1,
    "tokenInput": 0.2,
    "tokenInputCached": 0.02,
    "tokenOutput": 1.5,
    "contextWindow": "256k",
    "maxContextWindow": "-",
    "thinking": true,
    "isMax": false,
    "badges": [],
    "hidden": false,
    "notes": [],
    "docs": "https://docs.x.ai/docs/models#models-and-pricing"
  }];
  const calculateRequestsFromCost = costPerMillionTokens => {
    const margin = 0.2;
    let reqs = costPerMillionTokens / 0.04 * (1 + margin);
    reqs = Number(reqs).toFixed(2);
    reqs = parseFloat(reqs).toString();
    return <><span className="font-medium">{reqs}</span></>;
  };
  const parseContextWindow = contextString => {
    if (!contextString) return 0;
    const value = parseFloat(contextString);
    const unit = contextString.slice(-1).toUpperCase();
    if (unit === 'k') {
      return value * 1000;
    } else if (unit === 'M') {
      return value * 1000000;
    } else {
      return value;
    }
  };
  const [showHidden, setShowHidden] = useState(false);
  const [sortConfig, setSortConfig] = useState({
    key: 'name',
    direction: 'ascending'
  });
  const requestSort = key => {
    let direction = 'ascending';
    if (sortConfig.key === key && sortConfig.direction === 'ascending') {
      direction = 'descending';
    }
    setSortConfig({
      key,
      direction
    });
  };
  const getSortIndicator = key => {
    const isActive = sortConfig.key === key;
    return <span className="inline-flex w-4 h-4 ml-0.5" style={{
      transform: "translateY(4px)"
    }}>
        {isActive ? sortConfig.direction === "ascending" ? <svg className="w-full h-full" viewBox="0 0 24 24" fill="currentColor">
              <path d="M7.41 15.41L12 10.83l4.59 4.58L18 14l-6-6-6 6z"></path>
            </svg> : <svg className="w-full h-full" viewBox="0 0 24 24" fill="currentColor">
              <path d="M7.41 8.59L12 13.17l4.59-4.58L18 10l-6 6-6-6 1.41-1.41z"></path>
            </svg> : <svg className="w-full h-full opacity-0" viewBox="0 0 24 24">
            <path d="M7.41 15.41L12 10.83l4.59 4.58L18 14l-6-6-6 6z"></path>
          </svg>}
      </span>;
  };
  const Badge = ({badge, style}) => {
    return <span key={badge} style={style} className="flex-inline capitalize items-center px-1 py-0.5 rounded-sm text-xs font-medium">{badge}</span>;
  };
  const renderBadge = badge => {
    const badgeLower = badge.toLowerCase();
    const badgeStyles = {
      new: {
        backgroundColor: 'rgb(219 234 254)',
        color: 'rgb(30 64 175)',
        darkBackgroundColor: 'rgb(30 58 138)',
        darkColor: 'rgb(191 219 254)'
      },
      trending: {
        backgroundColor: 'rgb(243 232 255)',
        color: 'rgb(107 33 168)',
        darkBackgroundColor: 'rgb(88 28 135)',
        darkColor: 'rgb(233 213 255)'
      },
      recommended: {
        backgroundColor: 'rgb(197, 41, 176)',
        color: 'rgb(255 255 255)',
        darkBackgroundColor: 'rgb(197, 53, 101)',
        darkColor: 'rgb(255 255 255)'
      }
    };
    const style = badgeStyles[badgeLower] || ({
      backgroundColor: 'rgb(229 229 229)',
      color: 'rgb(38 38 38)',
      darkBackgroundColor: 'rgb(64 64 64)',
      darkColor: 'rgb(229 229 229)'
    });
    return <Badge badge={badge} style={{
      backgroundColor: style.backgroundColor,
      color: style.color,
      '@media (prefers-color-scheme: dark)': {
        backgroundColor: style.darkBackgroundColor,
        color: style.darkColor
      }
    }} />;
  };
  const getProviderSymbol = provider => {
    const urlPrefix = "https://mintlify.s3.us-west-1.amazonaws.com/cursor";
    switch (provider.toLowerCase()) {
      case 'openai':
        return <Tooltip tip="OpenAI">
          <img src={`${urlPrefix}/images/providers/openai-dark.svg`} alt="OpenAI" className="w-4 h-4 dark:hidden" />
          <img src={`${urlPrefix}/images/providers/openai-light.svg`} alt="OpenAI" className="w-4 h-4 hidden dark:block" />
        </Tooltip>;
      case 'anthropic':
        return <Tooltip tip="Anthropic">
          <img src={`${urlPrefix}/images/providers/anthropic-dark.svg`} alt="Anthropic" className="w-4 h-4 dark:hidden" />
          <img src={`${urlPrefix}/images/providers/anthropic-light.svg`} alt="Anthropic" className="w-4 h-4 hidden dark:block" />
        </Tooltip>;
      case 'google':
        return <Tooltip tip="Google">
          <img src={`${urlPrefix}/images/providers/google.svg`} alt="Google" className="w-4 h-4" />
        </Tooltip>;
      case 'deepseek':
        return <Tooltip tip="DeepSeek">
          <img src={`${urlPrefix}/images/providers/deepseek.png`} alt="DeepSeek" className="w-4 h-4" />
        </Tooltip>;
      case 'xai':
        return <Tooltip tip="xAI">
          <img src={`${urlPrefix}/images/providers/xai-dark.svg`} alt="xAI" className="w-4 h-4 dark:hidden" />
          <img src={`${urlPrefix}/images/providers/xai-light.svg`} alt="xAI" className="w-4 h-4 hidden dark:block" />
        </Tooltip>;
      case 'cursor':
        return <Tooltip tip="Cursor"><img src={`${urlPrefix}/images/providers/cursor.png`} alt="Cursor" className="w-4 h-4" /></Tooltip>;
      default:
        return <Tooltip tip={provider}><span>{provider[0]}</span></Tooltip>;
    }
  };
  const renderTableContent = isMaxMode => {
    const COLUMNS = {
      NAME: {
        id: 'name',
        style: {
          minWidth: '200px'
        },
        label: 'Name',
        sortable: true,
        render: model => {
          const isMaxOnly = !isMaxMode && model.isMax === "only";
          return <div className="flex items-center flex-wrap gap-1">
              <span className="mr-1 flex items-center">{getProviderSymbol(model.provider)}</span>
              {model.name}
              {model.notes.length > 0 && !isMaxOnly && <Tooltip tip={model.notes.join(', ')}>
                  <span className="relative inline-block ml-2">
                    <span className="w-4 h-4 bg-neutral-100 dark:bg-neutral-700 rounded-full inline-flex items-center justify-center text-xs font-bold text-neutral-600 dark:text-neutral-300">
                      <Icon icon="info" size={8} />
                    </span>
                  </span>
                </Tooltip>}
            </div>;
        }
      },
      CONTEXT: {
        id: 'contextWindow',
        label: 'Default Context',
        sortable: true,
        render: model => {
          if (model.isMax === "only") {
            return '-';
          }
          return <span>{model.contextWindow}</span>;
        }
      },
      MAX_CONTEXT: {
        id: 'maxContextWindow',
        label: 'Max Mode',
        sortable: true,
        render: model => {
          return <span>{model.maxContextWindow}</span>;
        }
      },
      TOKEN_INPUT: {
        id: 'tokenInput',
        label: `Input (MTok)`,
        tooltip: "Requests / 1M input tokens",
        render: model => model.isMax && model.hasOwnProperty('tokenInput') ? calculateRequestsFromCost(model.tokenInput) : ""
      },
      TOKEN_INPUT_CACHED: {
        id: 'tokenInputCached',
        label: `Cached Input (MTok)`,
        tooltip: "Requests / 1M cached input tokens",
        render: model => model.isMax && model.hasOwnProperty("tokenInputCached") ? calculateRequestsFromCost(model.tokenInputCached) : ""
      },
      TOKEN_OUTPUT: {
        id: 'tokenOutput',
        label: `Output (MTok)`,
        tooltip: "Requests / 1M output tokens",
        render: model => model.isMax ? calculateRequestsFromCost(model.tokenOutput) : ""
      },
      COST: {
        id: 'requests',
        label: `Cost (requests/message)`,
        sortable: true,
        tooltip: 'How resources are counted towards your quota',
        render: model => {
          if (isMaxMode && model.isMax) {
            return null;
          }
          if (!isMaxMode && model.isMax === "only") {
            return null;
          }
          return <>{model.requests === 0 ? 'Free' : `${Number(Number(model.requests).toFixed(2))}`}</>;
        }
      },
      CAPABILITIES: {
        id: 'capabilities',
        label: 'Capabilities',
        sortable: true,
        tooltip: 'Capabilities are the features and functionalities that an AI model can perform. These capabilities are determined by the model\'s design and training data.',
        render: model => {
          return <div className="grid grid-cols-3 gap-1 max-w-16">
              {model.isAgent ? <Tooltip tip="Agent: Can use tools">
                <svg width="14" height="7" viewBox="0 0 14 7" fill="none" className="text-black dark:text-white">
                  <path d="M0.432617 3.47461C0.432617 2.85938 0.55306 2.32389 0.793945 1.86816C1.03809 1.41243 1.37826 1.05924 1.81445 0.808594C2.25065 0.554688 2.75521 0.427734 3.32812 0.427734C3.78711 0.427734 4.2168 0.527018 4.61719 0.725586C5.02083 0.920898 5.42936 1.21712 5.84277 1.61426L7 2.73242L8.15723 1.61426C8.57064 1.21712 8.97917 0.920898 9.38281 0.725586C9.78646 0.527018 10.2161 0.427734 10.6719 0.427734C11.2448 0.427734 11.7493 0.554688 12.1855 0.808594C12.6217 1.05924 12.9603 1.41243 13.2012 1.86816C13.4453 2.32389 13.5674 2.85938 13.5674 3.47461C13.5674 4.08984 13.4453 4.62533 13.2012 5.08105C12.9603 5.53678 12.6217 5.8916 12.1855 6.14551C11.7493 6.39616 11.2448 6.52148 10.6719 6.52148C10.2161 6.52148 9.78646 6.42383 9.38281 6.22852C8.97917 6.02995 8.57064 5.7321 8.15723 5.33496L7 4.2168L5.84277 5.33496C5.42936 5.7321 5.02083 6.02995 4.61719 6.22852C4.2168 6.42383 3.78711 6.52148 3.32812 6.52148C2.75521 6.52148 2.25065 6.39616 1.81445 6.14551C1.37826 5.8916 1.03809 5.53678 0.793945 5.08105C0.55306 4.62533 0.432617 4.08984 0.432617 3.47461ZM1.52637 3.47461C1.52637 3.86849 1.60124 4.21354 1.75098 4.50977C1.90397 4.80273 2.11556 5.02897 2.38574 5.18848C2.65592 5.34798 2.97005 5.42773 3.32812 5.42773C3.6276 5.42773 3.91732 5.35449 4.19727 5.20801C4.47721 5.06152 4.77018 4.84505 5.07617 4.55859L6.2334 3.47461L5.08105 2.39062C4.77181 2.10417 4.47721 1.8877 4.19727 1.74121C3.91732 1.59473 3.6276 1.52148 3.32812 1.52148C2.97005 1.52148 2.65592 1.60124 2.38574 1.76074C2.11556 1.92025 1.90397 2.14648 1.75098 2.43945C1.60124 2.73242 1.52637 3.07747 1.52637 3.47461ZM7.7666 3.47461L8.92383 4.55859C9.22982 4.84505 9.52279 5.06152 9.80273 5.20801C10.0827 5.35449 10.3724 5.42773 10.6719 5.42773C11.0299 5.42773 11.3441 5.34798 11.6143 5.18848C11.8844 5.02897 12.0944 4.80273 12.2441 4.50977C12.3971 4.21354 12.4736 3.86849 12.4736 3.47461C12.4736 3.07747 12.3971 2.73242 12.2441 2.43945C12.0944 2.14648 11.8844 1.92025 11.6143 1.76074C11.3441 1.60124 11.0299 1.52148 10.6719 1.52148C10.3724 1.52148 10.0827 1.59473 9.80273 1.74121C9.52279 1.8877 9.22819 2.10417 8.91895 2.39062L7.7666 3.47461Z" fill="currentColor" />
                  <path d="M0.432617 3.47461C0.432617 2.85938 0.55306 2.32389 0.793945 1.86816C1.03809 1.41243 1.37826 1.05924 1.81445 0.808594C2.25065 0.554688 2.75521 0.427734 3.32812 0.427734C3.78711 0.427734 4.2168 0.527018 4.61719 0.725586C5.02083 0.920898 5.42936 1.21712 5.84277 1.61426L7 2.73242L8.15723 1.61426C8.57064 1.21712 8.97917 0.920898 9.38281 0.725586C9.78646 0.527018 10.2161 0.427734 10.6719 0.427734C11.2448 0.427734 11.7493 0.554688 12.1855 0.808594C12.6217 1.05924 12.9603 1.41243 13.2012 1.86816C13.4453 2.32389 13.5674 2.85938 13.5674 3.47461C13.5674 4.08984 13.4453 4.62533 13.2012 5.08105C12.9603 5.53678 12.6217 5.8916 12.1855 6.14551C11.7493 6.39616 11.2448 6.52148 10.6719 6.52148C10.2161 6.52148 9.78646 6.42383 9.38281 6.22852C8.97917 6.02995 8.57064 5.7321 8.15723 5.33496L7 4.2168L5.84277 5.33496C5.42936 5.7321 5.02083 6.02995 4.61719 6.22852C4.2168 6.42383 3.78711 6.52148 3.32812 6.52148C2.75521 6.52148 2.25065 6.39616 1.81445 6.14551C1.37826 5.8916 1.03809 5.53678 0.793945 5.08105C0.55306 4.62533 0.432617 4.08984 0.432617 3.47461ZM1.52637 3.47461C1.52637 3.86849 1.60124 4.21354 1.75098 4.50977C1.90397 4.80273 2.11556 5.02897 2.38574 5.18848C2.65592 5.34798 2.97005 5.42773 3.32812 5.42773C3.6276 5.42773 3.91732 5.35449 4.19727 5.20801C4.47721 5.06152 4.77018 4.84505 5.07617 4.55859L6.2334 3.47461L5.08105 2.39062C4.77181 2.10417 4.47721 1.8877 4.19727 1.74121C3.91732 1.59473 3.6276 1.52148 3.32812 1.52148C2.97005 1.52148 2.65592 1.60124 2.38574 1.76074C2.11556 1.92025 1.90397 2.14648 1.75098 2.43945C1.60124 2.73242 1.52637 3.07747 1.52637 3.47461ZM7.7666 3.47461L8.92383 4.55859C9.22982 4.84505 9.52279 5.06152 9.80273 5.20801C10.0827 5.35449 10.3724 5.42773 10.6719 5.42773C11.0299 5.42773 11.3441 5.34798 11.6143 5.18848C11.8844 5.02897 12.0944 4.80273 12.2441 4.50977C12.3971 4.21354 12.4736 3.86849 12.4736 3.47461C12.4736 3.07747 12.3971 2.73242 12.2441 2.43945C12.0944 2.14648 11.8844 1.92025 11.6143 1.76074C11.3441 1.60124 11.0299 1.52148 10.6719 1.52148C10.3724 1.52148 10.0827 1.59473 9.80273 1.74121C9.52279 1.8877 9.22819 2.10417 8.91895 2.39062L7.7666 3.47461Z" fill="currentColor" />
                </svg>
              </Tooltip> : <span />}
              {model.thinking ? <Tooltip tip="Thinking: Uses reasoning tokens">
                <svg width="12" height="10" viewBox="0 0 12 10" fill="none" className="text-black dark:text-white">
                  <path opacity="0.65" d="M0.503906 4.95312C0.503906 4.64583 0.55599 4.35156 0.660156 4.07031C0.764323 3.78646 0.91276 3.52995 1.10547 3.30078C1.29818 3.06901 1.52865 2.8776 1.79688 2.72656L2.25781 3.49609C1.98958 3.64193 1.77865 3.84505 1.625 4.10547C1.47396 4.36328 1.39844 4.64453 1.39844 4.94922C1.39844 5.28516 1.47005 5.57812 1.61328 5.82812C1.75911 6.07552 1.96224 6.26823 2.22266 6.40625C2.48307 6.54427 2.78776 6.61328 3.13672 6.61328C3.47526 6.61328 3.77083 6.55469 4.02344 6.4375C4.27604 6.31771 4.47266 6.15104 4.61328 5.9375C4.75391 5.72396 4.82422 5.47396 4.82422 5.1875C4.82422 5.05469 4.80599 4.9375 4.76953 4.83594C4.73307 4.73438 4.67969 4.65104 4.60938 4.58594C4.52083 4.5026 4.45052 4.4375 4.39844 4.39062C4.34635 4.34375 4.30859 4.29688 4.28516 4.25C4.26172 4.20052 4.25 4.13542 4.25 4.05469C4.25 3.9401 4.29036 3.84505 4.37109 3.76953C4.45443 3.69401 4.5599 3.65625 4.6875 3.65625C4.77083 3.65625 4.84635 3.67188 4.91406 3.70312C4.98438 3.73177 5.0638 3.78385 5.15234 3.85938C5.32943 4.00781 5.46484 4.19792 5.55859 4.42969C5.65495 4.65885 5.70312 4.91667 5.70312 5.20312C5.70312 5.66146 5.59505 6.0638 5.37891 6.41016C5.16536 6.75391 4.86458 7.02214 4.47656 7.21484C4.08854 7.40495 3.63672 7.5 3.12109 7.5C2.60026 7.5 2.14323 7.39453 1.75 7.18359C1.35677 6.97005 1.05078 6.67318 0.832031 6.29297C0.613281 5.91016 0.503906 5.46354 0.503906 4.95312ZM1.55859 2.8125C1.55859 2.47917 1.64062 2.1849 1.80469 1.92969C1.96875 1.67188 2.1901 1.47005 2.46875 1.32422C2.75 1.17578 3.0638 1.10156 3.41016 1.10156C3.65495 1.10156 3.89974 1.14323 4.14453 1.22656C4.39193 1.3099 4.6224 1.4349 4.83594 1.60156L4.30859 2.33203C4.16797 2.21745 4.02083 2.13281 3.86719 2.07812C3.71354 2.02083 3.55859 1.99219 3.40234 1.99219C3.22005 1.99219 3.05729 2.02734 2.91406 2.09766C2.77083 2.16797 2.65755 2.26432 2.57422 2.38672C2.49349 2.50911 2.45312 2.65104 2.45312 2.8125C2.45312 2.96094 2.48568 3.09115 2.55078 3.20312C2.61589 3.3151 2.70964 3.40365 2.83203 3.46875C2.95443 3.53125 3.09896 3.5625 3.26562 3.5625C3.38802 3.5625 3.49219 3.60677 3.57812 3.69531C3.66667 3.78125 3.71094 3.88542 3.71094 4.00781C3.71094 4.13021 3.66667 4.23568 3.57812 4.32422C3.49219 4.41016 3.38802 4.45312 3.26562 4.45312C2.92188 4.45312 2.6224 4.38411 2.36719 4.24609C2.11198 4.10807 1.91276 3.91536 1.76953 3.66797C1.62891 3.42057 1.55859 3.13542 1.55859 2.8125ZM3.71875 2.66016C3.72656 2.26172 3.82292 1.91016 4.00781 1.60547C4.19271 1.30078 4.44141 1.0625 4.75391 0.890625C5.06901 0.71875 5.42318 0.632812 5.81641 0.632812C6.09505 0.632812 6.35938 0.6875 6.60938 0.796875C6.85938 0.903646 7.07161 1.05208 7.24609 1.24219C7.29036 1.23438 7.33203 1.22786 7.37109 1.22266C7.41276 1.21745 7.44922 1.21484 7.48047 1.21484C7.76172 1.21484 8.02734 1.26823 8.27734 1.375C8.52734 1.48177 8.7474 1.63281 8.9375 1.82812C9.1276 2.02083 9.27604 2.25 9.38281 2.51562C9.49219 2.77865 9.54688 3.06641 9.54688 3.37891C9.54688 3.77474 9.48568 4.10677 9.36328 4.375C9.24349 4.64062 9.08854 4.86458 8.89844 5.04688C8.70833 5.22917 8.50651 5.38672 8.29297 5.51953C8.07943 5.65234 7.8776 5.77995 7.6875 5.90234C7.4974 6.02214 7.34115 6.15495 7.21875 6.30078C7.09896 6.44661 7.03906 6.625 7.03906 6.83594C7.03906 7.09375 7.14323 7.28776 7.35156 7.41797C7.5625 7.54557 7.84766 7.60938 8.20703 7.60938C8.27734 7.60938 8.33724 7.60938 8.38672 7.60938C8.4362 7.60677 8.47917 7.60547 8.51562 7.60547C8.57031 7.60547 8.61328 7.62109 8.64453 7.65234C8.67839 7.68359 8.69531 7.72526 8.69531 7.77734C8.69531 7.94922 8.71615 8.11589 8.75781 8.27734C8.80208 8.4388 8.86719 8.57031 8.95312 8.67188C9.03906 8.77604 9.14453 8.82812 9.26953 8.82812C9.39714 8.82812 9.51562 8.76562 9.625 8.64062C9.73438 8.51823 9.82292 8.33984 9.89062 8.10547C9.95833 7.87109 9.99219 7.58854 9.99219 7.25781C9.99219 7.04167 9.98307 6.84766 9.96484 6.67578C9.94922 6.5013 9.93099 6.32552 9.91016 6.14844L10.7969 5.90625C10.8203 6.09115 10.8398 6.28776 10.8555 6.49609C10.8711 6.70443 10.8789 6.96224 10.8789 7.26953C10.8789 7.60547 10.8424 7.92188 10.7695 8.21875C10.6966 8.51562 10.5911 8.77604 10.4531 9C10.3151 9.22656 10.1458 9.40365 9.94531 9.53125C9.7474 9.65885 9.52214 9.72266 9.26953 9.72266C9.04297 9.72266 8.83984 9.66276 8.66016 9.54297C8.48047 9.42318 8.33073 9.26302 8.21094 9.0625C8.09375 8.86458 8.01172 8.64583 7.96484 8.40625C7.9362 8.40885 7.90625 8.41016 7.875 8.41016C7.84635 8.41276 7.81771 8.41406 7.78906 8.41406C7.46094 8.41406 7.16927 8.35026 6.91406 8.22266C6.65885 8.09766 6.45833 7.92057 6.3125 7.69141C6.16927 7.45964 6.09766 7.1888 6.09766 6.87891C6.09766 6.56641 6.15885 6.30339 6.28125 6.08984C6.40625 5.8737 6.5651 5.6875 6.75781 5.53125C6.95312 5.375 7.15885 5.22917 7.375 5.09375C7.59375 4.95573 7.79948 4.8112 7.99219 4.66016C8.1875 4.50911 8.34635 4.33203 8.46875 4.12891C8.59375 3.92318 8.65625 3.67318 8.65625 3.37891C8.65625 3.12891 8.59635 2.91016 8.47656 2.72266C8.35938 2.53255 8.20443 2.38411 8.01172 2.27734C7.81901 2.16797 7.61068 2.11198 7.38672 2.10938C7.32161 2.10938 7.25651 2.11328 7.19141 2.12109C7.12891 2.12891 7.06901 2.14062 7.01172 2.15625C6.94922 2.16927 6.89583 2.17057 6.85156 2.16016C6.80729 2.14714 6.77083 2.11458 6.74219 2.0625C6.66927 1.90104 6.54818 1.77083 6.37891 1.67188C6.20964 1.57292 6.02083 1.52344 5.8125 1.52344C5.58333 1.52344 5.37891 1.57422 5.19922 1.67578C5.02214 1.77734 4.88021 1.91667 4.77344 2.09375C4.66927 2.26823 4.61328 2.47005 4.60547 2.69922C4.59766 2.83464 4.55208 2.94271 4.46875 3.02344C4.38802 3.10156 4.28516 3.14062 4.16016 3.14062C4.03255 3.14062 3.92578 3.09635 3.83984 3.00781C3.75651 2.91667 3.71615 2.80078 3.71875 2.66016ZM8.44141 6.63672C8.44141 6.51172 8.48568 6.40365 8.57422 6.3125C8.66536 6.22135 8.77344 6.18099 8.89844 6.19141C9.35677 6.23307 9.72266 6.14062 9.99609 5.91406C10.2695 5.6875 10.4062 5.36719 10.4062 4.95312C10.4062 4.5599 10.2839 4.2487 10.0391 4.01953C9.79688 3.79036 9.47656 3.6888 9.07812 3.71484L9.23047 2.80469C9.63932 2.84896 9.9987 2.96875 10.3086 3.16406C10.6211 3.35938 10.8646 3.61198 11.0391 3.92188C11.2135 4.22917 11.3008 4.57292 11.3008 4.95312C11.3008 5.41146 11.1992 5.80599 10.9961 6.13672C10.793 6.46745 10.5104 6.71615 10.1484 6.88281C9.78646 7.04948 9.36458 7.11589 8.88281 7.08203C8.75781 7.07422 8.65234 7.02865 8.56641 6.94531C8.48307 6.86198 8.44141 6.75911 8.44141 6.63672ZM6.51172 6.98438L6.82031 7.83984C6.61979 7.90755 6.41667 7.95833 6.21094 7.99219C6.00521 8.02344 5.80469 8.03906 5.60938 8.03906C5.19531 8.03906 4.8138 7.97135 4.46484 7.83594C4.11849 7.69792 3.83203 7.51302 3.60547 7.28125L4.46875 6.66406C4.54948 6.76562 4.65495 6.85286 4.78516 6.92578C4.91536 6.99609 5.0599 7.05078 5.21875 7.08984C5.38021 7.12891 5.54427 7.14844 5.71094 7.14844C5.84635 7.14844 5.98307 7.13411 6.12109 7.10547C6.26172 7.07682 6.39193 7.03646 6.51172 6.98438ZM5.00391 4.33203C5.38672 4.29557 5.66016 4.19661 5.82422 4.03516C5.98828 3.87109 6.0638 3.6276 6.05078 3.30469C6.04557 3.18229 6.08594 3.07812 6.17188 2.99219C6.25781 2.90365 6.36198 2.85938 6.48438 2.85938C6.60677 2.85938 6.71354 2.90365 6.80469 2.99219C6.89583 3.07812 6.94271 3.18229 6.94531 3.30469C6.95833 3.82552 6.82552 4.2474 6.54688 4.57031C6.27083 4.89323 5.85807 5.10156 5.30859 5.19531L5.00391 4.33203Z" fill="currentColor" />
                </svg>
              </Tooltip> : <span />}
            </div>;
        }
      },
      TRAIT: {
        id: 'trait',
        label: 'Description',
        sortable: true,
        tooltip: 'Model behaviour and goot to know',
        render: model => {
          return <div className="flex text-sm flex-col items-start gap-2"><p className="text-balance">{model.trait}</p>
            {!isPricing && model.badges.length === 1 && model.badges.map(renderBadge)}</div>;
        }
      }
    };
    const columns = isPricing ? isMaxMode ? [COLUMNS.NAME, COLUMNS.TOKEN_INPUT, COLUMNS.TOKEN_INPUT_CACHED, COLUMNS.TOKEN_OUTPUT] : [COLUMNS.NAME, COLUMNS.COST] : [COLUMNS.NAME, COLUMNS.CONTEXT, COLUMNS.MAX_CONTEXT, COLUMNS.CAPABILITIES];
    const sortedModels = useMemo(() => {
      let sortableItems = [...MODEL_LIST];
      sortableItems.sort((a, b) => {
        const hiddenA = a && a.hasOwnProperty('hidden');
        const hiddenB = b && b.hasOwnProperty('hidden');
        if (!showHidden) {
          if (hiddenA && !hiddenB) {
            return 1;
          }
          if (!hiddenA && hiddenB) {
            return -1;
          }
        }
        if (sortConfig !== null) {
          const aValue = a[sortConfig.key];
          const bValue = b[sortConfig.key];
          let comparison = 0;
          if (sortConfig.key === 'contextWindow' || sortConfig.key === 'maxContextWindow') {
            comparison = parseContextWindow(aValue) - parseContextWindow(bValue);
          } else if (typeof aValue === 'number' && typeof bValue === 'number') {
            comparison = aValue - bValue;
          } else if (typeof aValue === 'boolean' && typeof bValue === 'boolean') {
            comparison = aValue === bValue ? 0 : aValue ? -1 : 1;
          } else {
            comparison = String(aValue).toLowerCase().localeCompare(String(bValue).toLowerCase());
          }
          return sortConfig.direction === 'ascending' ? comparison : comparison * -1;
        }
        return 0;
      });
      return sortableItems.filter(model => {
        if (!showHidden && model.hasOwnProperty('hidden') && model.hidden === true) {
          return false;
        }
        return true;
      });
    }, [MODEL_LIST, sortConfig, showHidden]);
    return <div className="space-y-4">
        {isPricing && <div>
            {isMaxMode ? "Cost in requests per 1M token (MTok) from provider" : "Cost in requests per message"}
          </div>}

<div className="overflow-x-auto border dark:border-neutral-700 rounded-md">
          <table className="min-w-full divide-y divide-neutral-200 dark:divide-neutral-700 bg-white dark:bg-neutral-900">
            <thead className="dark:text-neutral-300">
              <tr>
                {columns.map(column => <th scope="col" style={column.style} className="px-4 py-3 text-left text-xs font-medium whitespace-nowrap cursor-pointer" onClick={() => requestSort(column.id)}>
                      {column.tooltip ? <Tooltip tip={column.tooltip}>
                          <span>{column.label}</span>
                        </Tooltip> : <span>{column.label}</span>}
                      {getSortIndicator(column.id)}
                    </th>)}
              </tr>
            </thead>
            <tbody className="divide-y divide-neutral-200 dark:divide-neutral-700">
              {sortedModels.map(model => <>
                  <tr key={model.id} className="align-text-top">
                    {columns.map(column => <td key={column.id} className="px-4 py-4 text-sm whitespace-nowrap text-black dark:text-white">
                        {column.id === 'name' && isPricing && model.subRows && model.subRows.some(subRow => subRow.hasOwnProperty(!isMaxMode ? 'requests' : 'tokenInput')) ? <div className="flex flex-col gap-1">
                            <div>{column.render ? column.render(model) : model[column.id]}</div>
                            {model.subRows.map(subRow => <div key={subRow.id} className="flex items-center gap-2 text-neutral-500 dark:text-neutral-400">
                                <span className="ml-6">{subRow.name}</span>
                                {subRow.notes?.map((note, index) => <Tooltip key={index} tip={note}>
                                    <span className="relative inline-block">
                                      <span className="w-4 h-4 bg-neutral-100 dark:bg-neutral-700 rounded-full inline-flex items-center justify-center text-xs font-bold text-neutral-600 dark:text-neutral-300">
                                        <Icon icon="info" size={8} />
                                      </span>
                                    </span>
                                  </Tooltip>)}
                              </div>)}
                          </div> : isPricing && model.subRows && model.subRows.some(subRow => subRow.hasOwnProperty(!isMaxMode ? 'requests' : 'tokenInput')) ? <div className="flex flex-col gap-1">
                            <div>{column.render ? column.render(model) : model[column.id]}</div>
                            {model.subRows.map(subRow => <div key={subRow.id} className="text-neutral-500 dark:text-neutral-400">
                                {column.render ? column.render(subRow) : subRow[column.id]}
                              </div>)}
                          </div> : column.render ? column.render(model) : model[column.id]}
                      </td>)}
                  </tr>
                </>)}
            </tbody>
          </table>
        </div>

<div className="flex justify-center mt-4">
          <button onClick={() => setShowHidden(!showHidden)} className="px-3 py-1 text-sm text-neutral-800 dark:text-neutral-200 rounded-full hover:bg-neutral-300 dark:hover:bg-neutral-600 transition-colors border border-neutral-200 dark:border-neutral-500">
            {showHidden ? "Hide models" : "Show more models"}
          </button>
        </div>
      </div>;
  };
  return <div className="not-prose space-y-4">{renderTableContent(false)}</div>;
};

Cursor supports all frontier coding models from all major model providers.

## Model pricing

Cursor [plans](/en/account/pricing) include usage at the model API rates. For example, \$20 of included usage on the Pro plan will be consumed based on your model selection and its price.

Usage limits are shown in editor based on your current consumption. To view all model API prices, view the provider documentation:

* [OpenAI Pricing](https://openai.com/api/pricing/)
* [Anthropic Pricing](https://www.anthropic.com/pricing#api)
* [Google Gemini Pricing](https://ai.google.dev/gemini-api/docs/pricing)
* [xAI Pricing](https://docs.x.ai/docs/models)

## Auto

## Context windows

A [context window](/en/guides/working-with-context) is the maximum span of tokens (text and code) an LLM can consider at once, including both the input prompt and output generated by the model.

Each chat in Cursor maintains its own context window. The more prompts, attached files, and responses included in a session, the larger the context window grows.

Learn more about [working with context](/en/guides/working-with-context) in Cursor.

## Max Mode

Normally, Cursor uses a context window of 200k tokens (\~15,000 lines of code). Max Mode extends the context window to the maximum available for a handful of models. This will be a bit slower and more expensive. It is most relevant for Gemini 2.5 Flash, Gemini 2.5 Pro, GPT 4.1, and Grok 4, which have context windows larger than 200k.

## FAQ

<AccordionGroup>
  <Accordion title="Where are models hosted?">
    Models are hosted on US-based infrastructure by the model's provider, a trusted partner, or Cursor directly.

When Privacy Mode is enabled, neither Cursor nor model providers store your data. All data is deleted after each request. For details see our [Privacy](/en/account/privacy), [Privacy Policy](https://cursor.com/privacy), and [Security](https://cursor.com/security) pages.
  </Accordion>
</AccordionGroup>

# API Keys
Source: https://docs.cursor.com/en/settings/api-keys

Bring your own LLM provider

Use your own API keys to send unlimited AI messages at your own cost. When configured, Cursor will use your API keys to call LLM providers directly.

To use your API key, go to `Cursor Settings` > `Models` and enter your API keys. Click **Verify**. Once validated, your API key is enabled.

<Warning>
  Custom API keys only work with standard chat models. Features requiring specialized models (like Tab Completion) will continue using Cursor's built-in models.
</Warning>

## Supported providers

* **OpenAI** - Standard, non-reasoning chat models only. The model picker will show the OpenAI models available.
* **Anthropic** - All Claude models available through the Anthropic API.
* **Google** - Gemini models available through the Google AI API.
* **Azure OpenAI** - Models deployed in your Azure OpenAI Service instance.
* **AWS Bedrock** - Use AWS access keys, secret keys, or IAM roles. Works with models available in your Bedrock configuration.

A unique external ID is generated after validating your Bedrock IAM role, which can be added to your IAM role trust policy for additional security.

## FAQ

<AccordionGroup>
  <Accordion title="Will my API key be stored or leave my device?">
    Your API key won't be stored but is sent to our server with every request. All requests are routed through our backend for final prompt building.
  </Accordion>
</AccordionGroup>

# Tab
Source: https://docs.cursor.com/en/tab/overview

Autocompletion with multi-line edits, cross-file suggestions, and context-aware code completions

Tab is a specialized Cursor model for autocompletion. The more you use it, the better it becomes as you inject intent by accepting <Kbd>Tab</Kbd> or rejecting <Kbd>Escape</Kbd> suggestions. With Tab, you can:

* Modify multiple lines at once
* Add import statements when missing
* Jump within and across files for coordinated edits
* Get suggestions based on recent changes, linter errors and accepted edits

## Suggestions

When adding text, completions appear as semi-opaque ghost text. When modifying existing code, it shows as a diff popup right of your current line.

Accept suggestions with <Kbd>Tab</Kbd>, reject with <Kbd>Escape</Kbd>, or accept word-by-word using <Kbd>Cmd+Arrow Right</Kbd>. Keep typing or press <Kbd>Escape</Kbd> to hide suggestions.

### Jump in file

Tab predicts your next editing location in the file and suggests jumps. After accepting an edit, press <Kbd>Tab</Kbd> again to jump to the next location.

### Jump across files

Tab predicts context-aware edits across files. A portal window appears at the bottom when a cross-file jump is suggested.

### Auto-import

In TypeScript and Python, Tab automatically adds import statements when missing. Use a method from another file and Tab suggests the import. Accepting adds it without disrupting your flow.

If auto-import isn't working:

* Ensure your project has the right language server or extensions
* Test with <Kbd>Cmd .</Kbd> to check if the import appears in *Quick Fix* suggestions

### Tab in Peek

Tab works in *Go to Definition* or *Go to Type Definition* peek views. Useful for modifying function signatures and fixing call sites.

In Vim, use with `gd` to jump to definitions, modify, and resolve references in one flow.

### Partial Accepts

Accept one word at a time with <Kbd>Cmd Right</Kbd>, or set your keybinding via `editor.action.inlineSuggest.acceptNextWord`. Enable in: `Cursor Settings` → `Tab`.

## Settings

### Toggling

Use the status bar (bottom-right) to:

* **Snooze**: Temporarily disable Tab for a chosen duration
* **Disable globally**: Disable Tab for all files
* **Disable for extensions**: Disable Tab for specific file extensions (e.g., markdown or JSON)

## FAQ

<AccordionGroup>
  <Accordion title="Tab gets in the way when writing comments, what can I do?">
    Disable Tab for comments by going to `Cursor Settings` → `Tab Completion` and unchecking **Trigger in comments**.
  </Accordion>

<Accordion title="Can I change the keyboard shortcut for Tab suggestions?">
    Remap accepting and rejecting suggestions to any key using `Accept Cursor Tab Suggestions` in `Keyboard Shortcuts` settings.
  </Accordion>

<Accordion title="How does Tab generate suggestions?">
    Cursor includes a small part of relevant code in the context window. The context is encrypted and sent to our backend. The backend decrypts and securely reads the context. Then, the Cursor Tab model predicts a code suggestion, and returns it to the client to display in editor.
  </Accordion>
</AccordionGroup>

# Developers
Source: https://docs.cursor.com/en/tools/developers

Generate install links for Tools & MCP Servers

export const McpInstallLinkGenerator = () => {
  const [config, setConfig] = useState("");
  const [error, setError] = useState("");
  const [showOverlay, setShowOverlay] = useState(null);
  const [extractedServerName, setExtractedServerName] = useState("");
  const debounceTimerRef = useRef(null);
  useEffect(() => {
    return () => {
      if (debounceTimerRef.current) {
        clearTimeout(debounceTimerRef.current);
      }
    };
  }, []);
  const handleConfigChange = e => {
    const configValue = e.target.value;
    setConfig(configValue);
    setError("");
    setExtractedServerName("");
    if (debounceTimerRef.current) {
      clearTimeout(debounceTimerRef.current);
    }
    if (configValue.trim()) {
      debounceTimerRef.current = setTimeout(() => {
        validateConfigWithValue(configValue);
      }, 500);
    }
  };
  const handleBlur = () => {
    if (debounceTimerRef.current) {
      clearTimeout(debounceTimerRef.current);
      debounceTimerRef.current = null;
    }
    if (config.trim()) {
      validateConfig();
    }
  };
  const validateConfig = () => {
    return validateConfigWithValue(config);
  };
  const validateConfigWithValue = configValue => {
    try {
      if (!configValue.trim()) {
        setError("");
        setExtractedServerName("");
        return false;
      }
      const parsedConfig = JSON.parse(configValue);
      if (typeof parsedConfig !== 'object' || parsedConfig === null) {
        throw new Error("Config must be a JSON object");
      }
      const configToUse = parsedConfig.mcpServers || parsedConfig;
      const serverName = Object.keys(configToUse)[0];
      if (!serverName) {
        throw new Error("No server configuration found");
      }
      const serverConfig = configToUse[serverName];
      if (typeof serverConfig !== 'object' || serverConfig === null) {
        throw new Error("Server config must be an object");
      }
      if (!serverConfig.command && !serverConfig.url) {
        throw new Error("Server config must have either 'command' or 'url' property");
      }
      if (serverConfig.command && typeof serverConfig.command !== 'string') {
        throw new Error("'command' must be a string");
      }
      if (serverConfig.url && typeof serverConfig.url !== 'string') {
        throw new Error("'url' must be a string");
      }
      if (serverConfig.args && !Array.isArray(serverConfig.args)) {
        throw new Error("'args' must be an array");
      }
      if (serverConfig.env && (typeof serverConfig.env !== 'object' || serverConfig.env === null)) {
        throw new Error("'env' must be an object");
      }
      setError("");
      setExtractedServerName(serverName);
      return true;
    } catch (e) {
      setError(e.message || "Invalid JSON configuration");
      setExtractedServerName("");
      return false;
    }
  };
  const INSTALL_BUTTON_IMAGE_URL = {
    DARK: "https://cursor.com/deeplink/mcp-install-dark.svg",
    LIGHT: "https://cursor.com/deeplink/mcp-install-light.svg"
  };
  const generateDeepLink = () => {
    if (!config.trim()) {
      setError("Config is required");
      return null;
    }
    try {
      const parsedConfig = JSON.parse(config);
      const configToUse = parsedConfig.mcpServers || parsedConfig;
      const serverName = Object.keys(configToUse)[0];
      let serverConfig = {
        ...configToUse[serverName]
      };
      if (serverConfig.command && serverConfig.args) {
        const argsString = serverConfig.args.join(" ");
        serverConfig.command = `${serverConfig.command} ${argsString}`;
        delete serverConfig.args;
      }
      const jsonString = JSON.stringify(serverConfig);
      const utf8Bytes = new TextEncoder().encode(jsonString);
      const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
      const safeBase64Config = base64Config.replace(/\+/g, '%2B');
      const protocol = window.location.hostname === 'localhost' ? 'cursor-dev' : 'cursor';
      return `${protocol}://anysphere.cursor-deeplink/mcp/install?name=${encodeURIComponent(serverName)}&config=${encodeURIComponent(safeBase64Config)}`;
    } catch (e) {
      setError(e.message || "Invalid JSON configuration");
      return null;
    }
  };
  const generateWebLink = () => {
    if (!config.trim()) {
      setError("Config is required");
      return null;
    }
    try {
      const parsedConfig = JSON.parse(config);
      const configToUse = parsedConfig.mcpServers || parsedConfig;
      const serverName = Object.keys(configToUse)[0];
      let serverConfig = {
        ...configToUse[serverName]
      };
      if (serverConfig.command && serverConfig.args) {
        const argsString = serverConfig.args.join(" ");
        serverConfig.command = `${serverConfig.command} ${argsString}`;
        delete serverConfig.args;
      }
      const jsonString = JSON.stringify(serverConfig);
      const utf8Bytes = new TextEncoder().encode(jsonString);
      const base64Config = btoa(Array.from(utf8Bytes).map(b => String.fromCharCode(b)).join(''));
      return `https://cursor.com/en/install-mcp?name=${encodeURIComponent(serverName)}&config=${encodeURIComponent(base64Config)}`;
    } catch (e) {
      setError(e.message || "Invalid JSON configuration");
      return null;
    }
  };
  const copyDeepLink = () => {
    const link = generateDeepLink();
    if (link) {
      navigator.clipboard.writeText(link);
      setShowOverlay('link');
      setTimeout(() => setShowOverlay(null), 1500);
    }
  };
  const copyWebLink = () => {
    const link = generateWebLink();
    if (link) {
      navigator.clipboard.writeText(link);
      setShowOverlay('weblink');
      setTimeout(() => setShowOverlay(null), 1500);
    }
  };
  const copyHtmlLink = theme => {
    const link = generateWebLink();
    if (link) {
      const imageUrl = INSTALL_BUTTON_IMAGE_URL[theme];
      const htmlLink = `<a href="${link}"><img src="${imageUrl}" alt="Add ${extractedServerName} MCP server to Cursor" height="32" /></a>`;
      navigator.clipboard.writeText(htmlLink);
      setShowOverlay(theme.toLowerCase());
      setTimeout(() => setShowOverlay(null), 1500);
    }
  };
  const copyMarkdownLink = theme => {
    const link = generateWebLink();
    if (link) {
      const imageUrl = INSTALL_BUTTON_IMAGE_URL[theme];
      const markdownLink = `[![Install MCP Server](${imageUrl})](${link})`;
      navigator.clipboard.writeText(markdownLink);
      setShowOverlay(`${theme.toLowerCase()}-md`);
      setTimeout(() => setShowOverlay(null), 1500);
    }
  };
  const copyJsxLink = theme => {
    const link = generateWebLink();
    if (link) {
      const imageUrl = INSTALL_BUTTON_IMAGE_URL[theme];
      const jsxLink = `<a href="${link}"><img src="${imageUrl}" alt="Add ${extractedServerName} MCP server to Cursor" height="32" /></a>`;
      navigator.clipboard.writeText(jsxLink);
      setShowOverlay(`${theme.toLowerCase()}-jsx`);
      setTimeout(() => setShowOverlay(null), 1500);
    }
  };
  return <div className="flex flex-col gap-3 w-full">
      <div className="relative">
        <textarea value={config} onChange={handleConfigChange} onBlur={handleBlur} placeholder={`{
  "postgres": {
    "command": "npx",
    "args": [
      "-y",
      "@modelcontextprotocol/server-postgres",
      "postgresql://localhost/mydb"
    ]
  }
}`} className="font-mono h-[250px] p-2 rounded border text-sm w-full rounded-t-xl border-neutral-300 dark:border-neutral-600 bg-white dark:bg-neutral-900 text-black dark:text-white placeholder-neutral-400 dark:placeholder-neutral-500" />
        {error && <div className="absolute bottom-5 px-2 py-1 text-red-500 rounded-lg left-3 bg-red-100 text-sm animate-in fade-in-0 slide-in-from-bottom-1 duration-200">
            {error}
          </div>}
      </div>
      <div className="flex flex-col sm:flex-row sm:items-center sm:justify-between gap-2 py-2 px-4 pr-2 border border-neutral-200 dark:border-neutral-500 rounded-lg transition-all duration-200">
        <div className="flex items-center gap-2 font-medium capitalize transition-colors duration-200">
          {extractedServerName.length > 0 ? <span className="animate-in fade-in-0 slide-in-from-left-2 duration-300">
              {extractedServerName}
            </span> : <span className="text-neutral-300">No server detected</span>}
        </div>
        <div className="flex gap-2">
          <div className="relative">
            <button onClick={copyDeepLink} disabled={!extractedServerName} className="border border-neutral-300 dark:border-neutral-600 text-neutral-700 dark:text-neutral-300 py-1 px-2 rounded-lg hover:bg-neutral-100 dark:hover:bg-neutral-800 bg-white dark:bg-neutral-900 text-sm transition-all duration-200 disabled:opacity-50 disabled:cursor-not-allowed">
              Copy deeplink
            </button>
            {showOverlay === 'link' && <div className="absolute inset-0 border border-neutral-300 bg-neutral-100 dark:bg-neutral-800 rounded-lg flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200">
                <Icon icon="check" size={16} />
                <span className="text-sm">Copied</span>
              </div>}
          </div>
          <div className="relative">
            <button onClick={copyWebLink} disabled={!extractedServerName} className="border border-neutral-300 dark:border-neutral-600 text-neutral-700 dark:text-neutral-300 py-1 px-2 rounded-lg hover:bg-neutral-100 dark:hover:bg-neutral-800 bg-white dark:bg-neutral-900 text-sm transition-all duration-200 disabled:opacity-50 disabled:cursor-not-allowed">
              Copy web link
            </button>
            {showOverlay === 'weblink' && <div className="absolute inset-0 border border-neutral-300 bg-neutral-100 dark:bg-neutral-800 rounded-lg flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200">
                <Icon icon="check" size={16} />
                <span className="text-sm">Copied</span>
              </div>}
          </div>
        </div>
      </div>
      <div className="flex flex-col gap-2 p-1">
        <div className="flex flex-col gap-2">
          <Tabs>
            <Tab title="Markdown">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyMarkdownLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 w-full max-w-40 sm:max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark-md' && <div className="top-2 right-2 bottom-2 left-2 absolute inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyMarkdownLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 w-full max-w-40 sm:max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light-md' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in README</p>
            </Tab>
            <Tab title="HTML">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyHtmlLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyHtmlLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in README</p>
            </Tab>
            <Tab title="JSX">
              <div className="flex flex-col sm:flex-row gap-2 justify-center pt-2">
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyJsxLink("DARK")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.DARK})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.DARK} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'dark-jsx' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} color="white" />
                        <span className="text-sm text-white">Copied</span>
                      </div>}
                  </div>
                </div>
                <div className="flex flex-col gap-1 items-center">
                  <div className="relative">
                    <button onClick={() => copyJsxLink("LIGHT")} className="p-2 border border-neutral-200 border-dashed rounded hover:opacity-90">
                      <div style={{
    backgroundImage: `url(${INSTALL_BUTTON_IMAGE_URL.LIGHT})`,
    backgroundSize: "cover",
    backgroundPosition: "center",
    backgroundRepeat: "no-repeat"
  }}>
                        <img src={INSTALL_BUTTON_IMAGE_URL.LIGHT} className="opacity-0 max-w-40 pointer-events-none" />
                      </div>
                    </button>
                    {showOverlay === 'light-jsx' && <div className="absolute top-2 right-2 bottom-2 left-2 inset-0 border border-neutral-200 border-dashed bg-white dark:bg-neutral-800 rounded flex items-center justify-center animate-in fade-in-0 zoom-in-95 duration-200 gap-2">
                        <Icon icon="check" size={16} className="text-black dark:text-white" />
                        <span className="text-sm text-black dark:text-white">Copied</span>
                      </div>}
                  </div>
                </div>
              </div>
              <p className="text-center mt-3 mb-2">Click to copy. Paste in JSX components</p>
            </Tab>

</Tabs>
        </div>
      </div>
    </div>;
};

# MCP Servers

MCP servers can be installed with Cursor deeplinks. It uses the same format as [`mcp.json`](/en/context/model-context-protocol.mdx) with a name and transport configuration.

Install links:

```
cursor://anysphere.cursor-deeplink/mcp/install?name=$NAME&config=$BASE64_ENCODED_CONFIG
```

## Generate install link

1. Get name and JSON configuration of server
2. `JSON.stringify` the configuration then base64 encode it
3. Replace `$NAME` and `$BASE64_ENCODED_CONFIG` with the name and encoded config

Helper for generating links:

## Example

Try this JSON in the MCP install link generator:

```json Single MCP server config theme={null}
{
  "postgres": {
    "command": "npx",
    "args": [
      "-y",
      "@modelcontextprotocol/server-postgres",
      "postgresql://localhost/mydb"
    ]
  }
}
```

Result:

<div className="full-width-table mcp-install-examples">
  | Format       | Example                                                                                                                                                                                                                                                                                                                                                        |
  | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
  | Text link    | [cursor://anysphere.curs...](cursor://anysphere.cursor-deeplink/mcp/install?name=postgres\&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=)                                                                                                                    |
  | Dark button  | <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=postgres&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=">  <img src="https://cursor.com/deeplink/mcp-install-dark.png" alt="Add postgres MCP server to Cursor" style={{ maxHeight: 32 }} /></a>  |
  | Light button | <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=postgres&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBtb2RlbGNvbnRleHRwcm90b2NvbC9zZXJ2ZXItcG9zdGdyZXMiLCJwb3N0Z3Jlc3FsOi8vbG9jYWxob3N0L215ZGIiXX0=">  <img src="https://cursor.com/deeplink/mcp-install-light.png" alt="Add postgres MCP server to Cursor" style={{ maxHeight: 32 }} /></a> |
</div>

## Install server

1. Click the link or paste into browser
2. Cursor prompts to install the server
3. Use the server in Cursor

# MCP Servers
Source: https://docs.cursor.com/en/tools/mcp

Explore and install MCP servers for Cursor

{hasConfirmation && <InstallModal isOpen={showModal} onClose={handleClose} deepLink={deepLink} server={server}>
          {prompt}
        </InstallModal>}
    </>;
};

Explore and install MCP servers for Cursor
Developers can [create an Add to Cursor button](/en/tools/developers#generate-install-link) and link to include in your developer documentation for easy installation in Cursor

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Notion</h2>

<p className="m-0">All-in-one workspace for notes, docs, and project management.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Figma</h2>

<p className="m-0">Design and collaboration platform for teams.</p>
    </div>

<McpInstallButton
      full
      server={{"name":"Figma","install":{"url":"http://127.0.0.1:3845/mcp"}}}
      prompt={<>
  Download the latest version of the <a href="https://www.figma.com/downloads/" target="_blank">Figma desktop app</a> to host the Figma MCP server. Check out the official <a href="https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Dev-Mode-MCP-Server" target="_blank">Figma setup guide</a> for installation instructions
    </>}
    />
  </Card>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Linear</h2>

<p className="m-0">Issue tracking and project management for development teams.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">GitHub</h2>

<p className="m-0">Version control and collaborative development platform.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Playwright</h2>

<p className="m-0">End-to-end browser testing.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Sentry</h2>

<p className="m-0">Error tracking and performance monitoring.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">DuckDB</h2>

<p className="m-0">In-process SQL OLAP database for local analytics.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Vercel</h2>

<p className="m-0">Manage projects and deployments on Vercel.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">GitLab</h2>

<p className="m-0">DevSecOps platform for code, CI/CD, and security.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Atlassian</h2>

<p className="m-0">Project management and collaboration tools including Jira and Confluence.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PostHog</h2>

<p className="m-0">Analytics, error tracking, and feature flags.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Stripe</h2>

<p className="m-0">Payment processing APIs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PayPal</h2>

<p className="m-0">Payment APIs.</p>
    </div>

<p className="m-0">dbt CLI, Semantic Layer, and Discovery API.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Browserbase</h2>

<p className="m-0">Headless browser sessions for agents.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Netlify</h2>

<p className="m-0">Build and deploy web projects.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Shopify</h2>

<p className="m-0">Shopify app development tools.</p>
    </div>

<p className="m-0">Vulnerability scanning of your codebase.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Zapier</h2>

<p className="m-0">Automate workflows between apps.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Heroku</h2>

<p className="m-0">Manage Heroku apps and resources.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Hugging Face</h2>

<p className="m-0">Access the Hugging Face Hub and Gradio MCP Servers.</p>
    </div>

<p className="m-0">Build and manage Wix sites.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Semgrep</h2>

<p className="m-0">Scan code for security vulnerabilities.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Plaid</h2>

<p className="m-0">Access financial account data.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Mercado Pago</h2>

<p className="m-0">Access Mercado Pago docs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">LaunchDarkly</h2>

<p className="m-0">Feature flags as a service.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Context7</h2>

<p className="m-0">Up-to-date code documentation.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">AWS Documentation</h2>

<p className="m-0">Access AWS documentation, search, and get recommendations.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Endor Labs</h2>

<p className="m-0">Security risk insights for code.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">InstantDB</h2>

<p className="m-0">Query and manage InstantDB.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Tinybird</h2>

<p className="m-0">Real-time analytics APIs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">MongoDB</h2>

<p className="m-0">Manage MongoDB data and deployments.</p>
    </div>

<p className="m-0">Manage Neon Postgres.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">SonarQube</h2>

<p className="m-0">Analyze code with SonarQube.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Socket</h2>

<p className="m-0">Analyze and secure dependencies.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Select Star</h2>

<p className="m-0">Data catalog, lineage, and context.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Pipedream</h2>

<p className="m-0">Connect to APIs and workflows.</p>
    </div>

<p className="m-0">Manage Auth0 resources.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">MS Learn Docs</h2>

<p className="m-0">Search Microsoft docs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Mercado Libre</h2>

<p className="m-0">Access Mercado Libre docs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Honeycomb</h2>

<p className="m-0">Query observability data and SLOs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">PagerDuty</h2>

<p className="m-0">Manage incidents and alerts.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Graphite</h2>

<p className="m-0">Create and manage stacked PRs.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Railway</h2>

<p className="m-0">Deploy apps, databases, and services.</p>
    </div>

<div className="flex items-baseline gap-2">
        <h2 className="m-0 text-neutral-900 dark:text-white">Postman</h2>

<p className="m-0">API collaboration and testing.</p>
    </div>

<div className="text-left mb-16 mt-4 max-w-[52rem] mx-auto flex flex-col gap-3 px-4 items-start text-sm">
  <p className="text-neutral-600 dark:text-neutral-400">
    This is a curated set of MCP tools for developers from official providers.
    If your company or tool provides an official MCP server and you want it
    included, please{" "}

{" "}

and we'll review it for inclusion.
  </p>
</div>

# Common Issues
Source: https://docs.cursor.com/en/troubleshooting/common-issues

Solutions for common problems and FAQs

Below are common issues and their solutions.

### Networking Issues

First, check your network connectivity. Go to `Cursor Settings` > `Network` and click `Run Diagnostics`. This will test your connection to Cursor's servers and help identify any network-related issues that might be affecting AI features, updates, or other online functionality.

Cursor relies on HTTP/2 for AI features because it handles streamed responses efficiently. If your network doesn't support HTTP/2, you may experience indexing failures and AI feature issues.

This occurs on corporate networks, VPNs, or when using proxies like Zscaler.

To resolve this, enable HTTP/1.1 fallback in app settings (not Cursor settings): press `CMD/CTRL + ,`, search for `HTTP/2`, then enable `Disable HTTP/2`. This forces HTTP/1.1 usage and resolves the issue.

We plan to add automatic detection and fallback.

### Resource Issues (CPU, RAM, etc.)

High CPU or RAM usage can slow your machine or trigger resource warnings.

While large codebases require more resources, high usage typically stems from extensions or settings issues.

<Note>
  If you are seeing a low RAM warning on **MacOS**, please note that there is a bug for some users that can show wildly incorrect values. If you are seeing this, please open the Activity Monitor and look at the "Memory" tab to see the correct memory usage.
</Note>

If you're experiencing high CPU or RAM usage, try these steps:

<AccordionGroup>
  <Accordion title="Check Your Extensions">
    Extensions can impact performance.

The Extension Monitor shows resource consumption for all your installed and built-in extensions.

Enable the extension monitor from `Settings` > `Application` > `Experimental` and toggle `Extension Monitor: Enabled`. This will prompt you to restart Cursor.

Open it: `Cmd/Ctrl + Shift + P` → `Developer: Open Extension Monitor`.

Cursor runs your extensions in one or more **extension hosts**. Usually, most of your extensions will run in the same extension host, which means an extension consuming a lot of CPU time can suffocate its neighboring extensions!

The Extension Monitor will show:

* Every long-lived process launched by an extension (MacOS and Linux only).
    * **% Ext Host**: The percentage of total extension host time consumed by this extension. Helps identify which extensions are using the most time relative to others.
    * **Max Blocking**: An extensions's longest continuous block of execution, per monitoring interval.
    * **% CPU**:
      * For extensions: The percentage of total CPU usage attributed to the extension's code.
      * For processes: The percentage of total CPU usage attributed to the launched process (MacOS and Linux only).
    * **Memory**:
      * For extensions: The amount of JS heap memory used by the extension's code (external allocations not counted).
      * For processes: The amount of system memory used by the launched process (MacOS and Linux only).

You can also test by running `cursor --disable-extensions` from the command line. If performance improves, re-enable extensions one by one to identify problematic ones.

Try Extension Bisect to identify problematic extensions. Read more [here](https://code.visualstudio.com/blogs/2021/02/16/extension-bisect#_welcome-extension-bisect). Note: this works best for immediate issues, not gradual performance degradation.
  </Accordion>

<Accordion title="Use the Process Explorer">
    The Process Explorer shows which processes consume resources.

Open it: Command Palette (`Cmd/Ctrl + Shift + P`) → `Developer: Open Process Explorer`.

Review processes under:

* **`extensionHost`**: Extension-related issues
    * **`ptyHost`**: Terminal resource consumption

The Process Explorer displays each terminal and its running commands for diagnosis.

For other high-usage processes, report to the [forum](https://forum.cursor.com/).
  </Accordion>

<Accordion title="Monitor System Resources">
    Use your operating system's monitoring tools to determine if the issue is Cursor-specific or system-wide.
  </Accordion>

<Accordion title="Testing a Minimal Installation">
    If issues persist, test a minimal Cursor installation.
  </Accordion>
</AccordionGroup>

## General FAQs

<AccordionGroup>
  <Accordion title="I see an update on the changelog but Cursor won't update">
    New updates use staged rollouts - randomly selected users receive updates first. Expect your update within a few days.
  </Accordion>

<Accordion title="I have issues with my GitHub login in Cursor / How do I log out of GitHub in Cursor?">
    Use `Sign Out of GitHub` from the command palette `Ctrl/⌘ + Shift + P`.
  </Accordion>

<Accordion title="I can't use GitHub Codespaces">
    GitHub Codespaces isn't supported yet.
  </Accordion>

<Accordion title="I have errors connecting to Remote SSH">
    SSH to Mac or Windows machines isn't supported. For other issues, report to the [forum](https://forum.cursor.com/) with logs.
  </Accordion>

<Accordion title="SSH Connection Problems on Windows">
    If you see "SSH is only supported in Microsoft versions of VS Code":

1. Uninstall Remote-SSH:
       * Open Extensions view (`Ctrl + Shift + X`)
       * Search "Remote-SSH"
       * Click gear icon → "Uninstall"

2. Install Anysphere Remote SSH:
       * Open Cursor marketplace
       * Search "SSH"
       * Install the Anysphere Remote SSH extension

3. After installation:
       * Close all VS Code instances with active SSH connections
       * Restart Cursor
       * Reconnect via SSH

Verify your SSH configuration and keys are properly set up.
  </Accordion>

<Accordion title="Cursor Tab and Inline Edit do not work behind my corporate proxy">
    Cursor Tab and Inline Edit use HTTP/2 for lower latency and resource usage. Some corporate proxies (e.g., Zscaler) block HTTP/2. Fix by setting `"cursor.general.disableHttp2": true` in settings (`Cmd/Ctrl + ,`, search `http2`).
  </Accordion>

<Accordion title="I just subscribed to Pro but I'm still on the free plan in the app">
    Log out and back in from Cursor Settings.
  </Accordion>

<Accordion title="When will my usage reset again?">
    Pro subscribers: Click `Manage Subscription` in the [Dashboard](https://cursor.com/dashboard) to view your renewal date.

Free users: Check your first Cursor email date. Usage resets monthly from that date.
  </Accordion>

<Accordion title="My Chat/Composer history disappeared after an update">
    Low disk space may cause Cursor to clear historical data during updates. To prevent this:

1. Maintain sufficient free disk space before updating
    2. Clean up unnecessary system files regularly
    3. Back up important conversations before updating
  </Accordion>

<Accordion title="How do I uninstall Cursor?">
    Follow [this guide](https://code.visualstudio.com/docs/setup/uninstall). Replace "VS Code" or "Code" with "Cursor", and ".vscode" with ".cursor".
  </Accordion>

<Accordion title="How do I delete my account?">
    Click `Delete Account` in the [Dashboard](https://cursor.com/dashboard). This permanently deletes your account and all associated data.
  </Accordion>

<Accordion title="How do I open Cursor from the command line?">
    Run `cursor` in your terminal. If the command is missing:

1. Open command palette `⌘⇧P`
    2. Type `install command`
    3. Select `Install 'cursor' command` (optionally install `code` command to override VS Code's)
  </Accordion>

<Accordion title="Unable to Sign In to Cursor">
    If clicking Sign In redirects to cursor.com but doesn't sign you in, disable your firewall or antivirus software - they may block the sign-in process.
  </Accordion>

<Accordion title="Suspicious Activity Message">
    Due to recent increased misuse of our system, your request may have been blocked as a security measure. Here's how to resolve this:

First, check your VPN. If you're using one, try turning it off, as VPNs can sometimes trigger our security systems.

If that doesn't resolve it, you can try:

* Creating a new chat
    * Waiting a bit and trying again later
    * Creating a new account using Google or GitHub authentication
    * Upgrading to Cursor Pro
  </Accordion>
</AccordionGroup>

# Getting a Request ID
Source: https://docs.cursor.com/en/troubleshooting/request-reporting

Find request IDs for technical support

When the Cursor team is investigating a technical issue, we may ask for a "request ID".

## What is a request ID?

A request ID uniquely identifies each request to Cursor in our internal systems.

Format example: `8f2a5b91-4d3e-47c6-9f12-5e8d94ca7d23`

## How do I find a request ID?

<Warning>
  Request IDs are limited when Privacy Mode is enabled. Disable Privacy Mode when reporting issues.

Note: Business plan users have Privacy Mode enabled by default by their organization's admin.
</Warning>

### Getting your current request ID

To report an issue with your current or recent conversation:

With the conversation open in the Chat sidebar, use the context menu (top right) and select `Copy Request ID`.

Send the copied request ID to us via the forum or email as requested.

### Getting a request ID from a previous action

Retrieve historical request IDs using the `Report AI Action` command:

1. Open command palette `⌘⇧P`
2. Type `Report AI Action`
3. Select the `Report AI Action` option

This displays your recent AI actions across Chat, CMD+K, and Apply.

Select the action by matching its time and feature. Copy the request ID and send it to us.

# Troubleshooting Guide
Source: https://docs.cursor.com/en/troubleshooting/troubleshooting-guide

Steps to fix issues and report bugs

Cursor issues can stem from extensions, app data, or system problems. Try these troubleshooting steps.

<CardGroup cols={1}>
  <Card horizontal title="Reporting an Issue" icon="bug" href="#reporting-an-issue">
    Steps to report an issue to the Cursor team
  </Card>
</CardGroup>

## Troubleshooting

<Steps>
  <Step title="Check network connectivity">
    First check if Cursor can connect to its services.

**Run network diagnostics:** Go to `Cursor Settings` > `Network` and click `Run Diagnostics`. This tests your connection to Cursor's servers and identifies network issues affecting AI features, updates, or other online functionality.

If diagnostics reveal connectivity issues, check firewall settings, proxy configuration, or network restrictions blocking Cursor's access.
  </Step>

<Step title="Clearing extension data">
    For extension issues:

**Disable all extensions temporarily:** Run `cursor --disable-extensions` from the command line. If issues resolve, re-enable extensions one by one to identify the problematic one.

**Reset extension data:** Uninstall and reinstall problematic extensions to reset their stored data. Check settings for extension configuration that persists after reinstallation.
  </Step>

<Step title="Clearing app data">
    <Warning>
      This deletes your app data, including extensions, themes, snippets, and installation-related data. Export your profile first to preserve this data.
    </Warning>

Cursor stores app data outside the app for restoration between updates and reinstallations.

To clear app data:

**Windows:** Run these commands in Command Prompt:

**MacOS:** Run `sudo rm -rf ~/Library/Application\ Support/Cursor` and `rm -f ~/.cursor.json` in Terminal.

**Linux:** Run `rm -rf ~/.cursor ~/.config/Cursor/` in Terminal.
  </Step>

<Step title="Uninstalling Cursor">
    To uninstall Cursor:

<CardGroup cols={1}>
      <Card horizontal title="Windows" icon="windows">
        Search "Add or Remove Programs" in Start Menu, find "Cursor", click "Uninstall".
      </Card>

<Card horizontal title="MacOS" icon="apple">
        Open Applications folder, right-click "Cursor", select "Move to Trash".
      </Card>

<Card horizontal title="Linux" icon="linux">
        **For .deb packages:** `sudo apt remove cursor`

**For .rpm packages:** `sudo dnf remove cursor` or `sudo yum remove cursor`

**For AppImage:** Delete the Cursor.appimage file from its location.
      </Card>
    </CardGroup>
  </Step>

<Step title="Reinstalling Cursor">
    Reinstall from the [Downloads page](https://www.cursor.com/downloads). Without cleared app data, Cursor restores to its previous state. Otherwise, you get a fresh install.
  </Step>
</Steps>

## Reporting an Issue

If these steps don't help, report to the [forum](https://forum.cursor.com/).

<Card horizontal title="Cursor Forum" icon="message" href="https://forum.cursor.com/">
  Report a bug or issue on the Cursor forum
</Card>

For quick resolution, provide:

<CardGroup cols={2}>
  <Card title="Screenshot of Issue" icon="image">
    Capture a screenshot, redact sensitive information.
  </Card>

<Card title="Steps to Reproduce" icon="list-check">
    Document exact steps to reproduce the issue.
  </Card>

<Card title="System Information" icon="computer">
    Get system info from:

<br />

`Cursor` > `Help` > `About`
  </Card>

<Card title="Request IDs" icon="shield-halved" href="/en/troubleshooting/request-reporting">
    Click to view our guide on gathering request IDs
  </Card>

<Card title="Console Errors" icon="bug">
    Check developer console: <br />
    `Developer: Toggle Developer Tools`
  </Card>

<Card title="Logs" icon="file-lines">
    Access logs: <br />
    `Developer: Open Logs Folder`
  </Card>
</CardGroup>

# Welcome
Source: https://docs.cursor.com/en/welcome

Learn about Cursor and how to get started

Cursor is an AI-powered code editor that understands your codebase and helps you code faster through natural language. Just describe what you want to build or change and Cursor will generate the code for you.

<CardGroup cols={3}>
  <Card title="Get started" icon="rocket" href="/en/get-started/installation">
    <div>
      Download, install, and start building with Cursor in minutes
    </div>
  </Card>

<Card title="Changelog" icon="sparkles" href="https://www.cursor.com/changelog">
    <div>
      Stay up to date with the latest features and improvements
    </div>
  </Card>

<Card title="CLI" icon="terminal" href="/en/cli/overview">
    <div>
      Use Cursor in your terminal
    </div>
  </Card>

<Card title="Concepts" icon="lightbulb" href="/en/get-started/concepts">
    <div>
      Understand core concepts and features that power Cursor
    </div>
  </Card>

<Card title="Models" icon="brain" href="/en/models">
    <div>
      Explore AI models available and how to select the right one
    </div>
  </Card>

<Card title="Guides" icon="book" href="/en/guides/working-with-context">
    <div>
      Learn best practices and workflows for different use cases
    </div>
  </Card>

<Card title="Downloads" icon="download" href="https://cursor.com/downloads" arrow>
    <div>
      Get Cursor for your computer
    </div>
  </Card>

<Card title="Forum" icon="message" href="https://forum.cursor.com">
    <div>
      For technical queries and to share experiences, visit our forum
    </div>
  </Card>

<Card title="Support" icon="headset" href="mailto:hi@cursor.com">
    <div>
      For account and billing questions, email our support team
    </div>
  </Card>
</CardGroup>

# Seguridad del Agente
Source: https://docs.cursor.com/es/account/agent-security

Consideraciones de seguridad para usar Cursor Agent

La inyección de prompts, las alucinaciones de IA y otros problemas pueden hacer que la IA se comporte de formas inesperadas y potencialmente maliciosas. Mientras seguimos trabajando para resolver la inyección de prompts a un nivel más fundamental, nuestra protección principal en los productos de Cursor son los rieles de seguridad sobre lo que un agente puede hacer, incluyendo exigir aprobación manual para acciones sensibles de forma predeterminada. El objetivo de este documento es explicar estos rieles de seguridad y qué pueden esperar los usuarios de ellos.

Todos los controles y comportamientos a continuación son nuestra configuración predeterminada y recomendada.

<div id="first-party-tool-calls">
  ## Llamadas a herramientas nativas
</div>

Cursor viene con herramientas integradas que le permiten al agente ayudarte a escribir código. Estas incluyen lectura de archivos, edición, ejecución de comandos de terminal, búsqueda de documentación en la web y otras.

Las herramientas de lectura no requieren aprobación (p. ej., leer archivos, buscar en el código). Puedes usar [.cursorignore](/es/context/ignore-files) para bloquear por completo el acceso del agente a archivos específicos, pero en caso contrario las lecturas generalmente están permitidas sin aprobación. Para acciones que conllevan riesgo de exfiltración de datos sensibles, requerimos aprobación explícita.

Modificar archivos dentro del espacio de trabajo actual no requiere aprobación explícita con algunas excepciones. Cuando un agente realiza cambios en archivos, estos se guardan inmediatamente en disco. Recomendamos ejecutar Cursor en espacios de trabajo con control de versiones, de modo que el contenido de los archivos pueda revertirse en cualquier momento. Requerimos aprobación explícita antes de cambiar archivos que modifiquen la configuración de nuestro IDE/CLI, como el archivo de configuración del espacio de trabajo del editor. Sin embargo, si tienes recarga automática al detectar cambios en archivos, ten en cuenta que los cambios del agente pueden activar ejecuciones automáticas antes de que hayas tenido oportunidad de revisarlos.

Cualquier comando de terminal sugerido por los agentes requiere aprobación de forma predeterminada. Recomendamos revisar cada comando antes de que el agente lo ejecute. Quienes acepten el riesgo pueden habilitar que el agente ejecute todos los comandos sin aprobación. Incluimos una [allowlist](/es/agent/tools) en Cursor, pero no la consideramos un control de seguridad. Algunas personas eligen permitir comandos específicos, pero es un sistema de mejor esfuerzo y pueden existir formas de eludirlo. No recomendamos "Run Everything", que omite cualquier allowlist configurada.

<div id="third-party-tool-calls">
  ## Llamadas a herramientas de terceros
</div>

Cursor permite conectar herramientas externas a través de [MCP](/es/context/mcp). Todas las conexiones MCP de terceros deben ser aprobadas explícitamente por el usuario. Una vez que el usuario aprueba un MCP, de forma predeterminada cada llamada de herramienta sugerida en Agent Mode para cualquier integración MCP externa debe ser aprobada explícitamente antes de su ejecución.

<div id="network-requests">
  ## Solicitudes de red
</div>

Un atacante podría usar las solicitudes de red para exfiltrar datos. Actualmente no admitimos que herramientas propias realicen solicitudes de red fuera de un conjunto muy selecto de hosts (p. ej., GitHub), la recuperación explícita de enlaces ni el uso de la búsqueda web más que con un conjunto selecto de proveedores. Con la configuración predeterminada se impiden las solicitudes de red arbitrarias por parte de agentes.

<div id="workspace-trust">
  ## Confianza del espacio de trabajo
</div>

El IDE de Cursor admite la función estándar de [workspace trust](https://code.visualstudio.com/docs/editing/workspaces/workspace-trust), que está *desactivada* de forma predeterminada. Workspace trust te muestra un aviso al abrir un nuevo espacio de trabajo para elegir entre modo normal o restringido. El modo restringido inutiliza la IA y otras funciones por las que normalmente usas Cursor. Te recomendamos usar otras herramientas, como un editor de texto básico, para trabajar con repos en los que no confíes.

Puedes habilitar workspace trust en la configuración de usuario siguiendo estos pasos:

1. Abre tu archivo de usuario settings.json
2. Agrega la siguiente configuración:
   ```json  theme={null}
   "security.workspace.trust.enabled": true
   ```

Esta configuración también puede aplicarse a nivel de organización mediante soluciones de Mobile Device Management (MDM).

<div id="responsible-disclosure">
  ## Divulgación responsable
</div>

Si crees que has encontrado una vulnerabilidad en Cursor, sigue la guía en nuestra página de seguridad de GitHub y envía el informe ahí. Si no puedes usar GitHub, también puedes escribirnos a [security@cursor.com](mailto:security@cursor.com).

Nos comprometemos a acusar recibo de los informes de vulnerabilidades en un plazo de 5 días hábiles y a resolverlos tan pronto como podamos. Publicaremos los resultados en forma de avisos de seguridad en nuestra página de seguridad de GitHub. Los incidentes críticos se comunicarán tanto en la página de seguridad de GitHub como por correo electrónico a todos los usuarios.

# Facturación
Source: https://docs.cursor.com/es/account/billing

Gestión de suscripciones, reembolsos y facturas de Cursor

<div id="how-do-i-access-billing-settings">
  ### How do I access billing settings?
</div>

Accede al portal de facturación desde el [Dashboard](https://cursor.com/dashboard) haciendo clic en "Billing" en tu dashboard. Se abrirá un portal seguro para todas las tareas de facturación.

<div id="what-are-cursors-billing-cycles">
  ### What are Cursor's billing cycles?
</div>

Los ciclos de facturación pueden ser mensuales o anuales y comienzan en la fecha de tu suscripción. Las cuentas de Teams se cobran por asiento, con prorrateo para miembros nuevos.

<div id="how-do-seats-work-for-teams-accounts">
  ### How do seats work for Teams accounts?
</div>

Las cuentas de Teams cobran por asiento (uno por cada miembro del equipo). Si agregas miembros a mitad de ciclo, solo se te cobra por el tiempo restante. Si un miembro ha usado créditos y lo quitas, su asiento permanece ocupado hasta el final del ciclo de facturación; no hay reembolsos prorrateados. Los administradores del equipo pueden gestionar los asientos desde el dashboard.

<div id="can-i-switch-between-monthly-and-annual-billing">
  ### Can I switch between monthly and annual billing?
</div>

¡Sí! Así es cómo:

**Pro plan**

1. Ve al [dashboard](https://cursor.com/dashboard) de Cursor
2. Haz clic en "Billing and Invoices" en la barra lateral izquierda para ir a la página de facturación
3. Haz clic en "Manage subscription"
4. Haz clic en "Update subscription"
5. Selecciona "Yearly" o "Monthly" y luego haz clic en "Continue"

**Teams plan**

1. Ve al [dashboard](https://cursor.com/dashboard) de Cursor
2. Haz clic en "Billing and Invoices" en la barra lateral izquierda para ir a la página de facturación
3. Haz clic en el botón "Upgrade Now" para cambiar a facturación anual

<Note>
  Solo puedes cambiar de facturación mensual a anual por tu cuenta. Para cambiar de anual a mensual, contáctanos en
  [hi@cursor.com](mailto:hi@cursor.com).
</Note>

<div id="where-can-i-find-my-invoices">
  ### Where can I find my invoices?
</div>

Encuentra todo tu historial de facturación en el portal de facturación. Puedes ver y descargar facturas actuales y pasadas.

<div id="can-i-get-invoices-automatically-emailed-to-me">
  ### Can I get invoices automatically emailed to me?
</div>

Las facturas deben descargarse manualmente desde el portal de facturación. Estamos desarrollando el envío automático por email. Podrás activar la opción cuando esté disponible.

<div id="how-do-i-update-my-billing-information">
  ### How do I update my billing information?
</div>

Actualiza el método de pago, el nombre de la empresa, la dirección y la información fiscal desde el portal de facturación. Usamos Stripe para transacciones seguras. Los cambios solo afectan a facturas futuras; no podemos modificar facturas anteriores.

<div id="how-do-i-cancel-my-subscription">
  ### How do I cancel my subscription?
</div>

Cancela tu suscripción desde la página "Billing and Invoices" haciendo clic en "Manage Subscription" y luego en el botón "Cancel subscription". Seguirás teniendo acceso hasta el final de tu período de facturación actual.

<div id="im-having-other-billing-issues-how-can-i-get-help">
  ### I'm having other billing issues. How can I get help?
</div>

Para consultas de facturación no cubiertas aquí, envía un email a [hi@cursor.com](mailto:hi@cursor.com) desde el email asociado a tu cuenta. Incluye los detalles de tu cuenta y tus inquietudes.

# Precios
Source: https://docs.cursor.com/es/account/pricing

Los planes de Cursor y sus precios

Podés probar Cursor gratis o comprar un plan individual o de equipo.

<div id="individual">
  ## Individual
</div>

Todos los planes individuales incluyen:

* Autocompletados en pestañas ilimitados
* Límites ampliados de uso de agentes en todos los modelos
* Acceso a Bugbot
* Acceso a Background Agents

Cada plan incluye uso facturado según los precios de inferencia del modelo [API prices](/es/models#model-pricing):

* Pro incluye \$20 de uso de API de agentes + uso adicional de bonificación
* Pro Plus incluye \$70 de uso de API de agentes + uso adicional de bonificación
* Ultra incluye \$400 de uso de API de agentes + uso adicional de bonificación

Trabajamos mucho para otorgar capacidad de bonificación adicional más allá del uso incluido garantizado. Como distintos modelos tienen diferentes costos de API, la selección de modelo afecta la salida de tokens y qué tan rápido se consume tu uso incluido. Puedes ver el uso y los desgloses de tokens en [tu panel](https://cursor.com/dashboard?tab=usage). Las notificaciones de límite se muestran de forma habitual en el editor.

<div id="how-much-usage-do-i-need">
  ### ¿Cuánto uso necesito?
</div>

Según nuestros datos de uso, puedes esperar los siguientes niveles:

* **Usuarios que usan Tab a diario**: Siempre se mantienen dentro de \$20
* **Usuarios con uso limitado de Agent**: A menudo se mantienen dentro de los \$20 incluidos
* **Usuarios que usan Agent a diario**: Normalmente $60–$100/mes de uso total
* **Usuarios avanzados (varios agents/automatización)**: A menudo \$200+/mes de uso total

Según nuestros datos de uso, los límites son aproximadamente equivalentes a lo siguiente para un *usuario mediano*:

* Pro: \~225 solicitudes de Sonnet 4, \~550 solicitudes de Gemini o \~500 solicitudes de GPT 5
* Pro+: \~675 solicitudes de Sonnet 4, \~1,650 solicitudes de Gemini o \~1,500 solicitudes de GPT 5
* Ultra: \~4,500 solicitudes de Sonnet 4, \~11,000 solicitudes de Gemini o \~10,000 solicitudes de GPT 5

<div id="what-happens-when-i-reach-my-limit">
  ### ¿Qué pasa cuando llego a mi límite?
</div>

Cuando superes tu uso mensual incluido, vas a recibir una notificación en el editor y podés elegir:

* **Agregar uso bajo demanda**: Seguí usando Cursor a las mismas tarifas de API con facturación por consumo
* **Actualizar tu plan**: Pasá a un nivel superior para tener más uso incluido

El uso bajo demanda se factura mensualmente a las mismas tarifas que tu uso incluido. Las solicitudes nunca se degradan en calidad ni velocidad.

<div id="teams">
  ## Teams
</div>

Hay dos planes para equipos: Teams (\$40/usuario/mes) y Enterprise (Personalizado).

Los planes de Teams incluyen funciones adicionales como:

* Aplicación del Privacy Mode
* Panel de administración con estadísticas de uso
* Facturación centralizada del equipo
* SSO con SAML/OIDC

Recomendamos Teams para cualquier cliente que esté cómodo autoatendiéndose. Recomendamos [Enterprise](/es/contact-sales) para clientes que necesiten soporte prioritario, uso compartido, facturación, SCIM o controles de seguridad avanzados.

Obtén más información sobre los [precios de Teams](/es/account/teams/pricing).

Activar Auto permite que Cursor elija el modelo premium más adecuado para la tarea inmediata y con la mayor fiabilidad según la demanda actual. Esta función puede detectar una degradación en el rendimiento de salida y cambiar de modelo automáticamente para solucionarlo.

<Note>Hemos invertido mucho en la calidad y el rendimiento general de Auto. A partir de tu próxima renovación de facturación después del 15 de septiembre, Auto consumirá a las siguientes tarifas de API.</Note>

* **Input + Cache Write**: \$1.25 por 1M de tokens
* **Output**: \$6.00 por 1M de tokens
* **Cache Read**: \$0.25 por 1M de tokens

Tanto el editor como el panel mostrarán tu consumo, que incluye Auto. Si prefieres seleccionar un modelo directamente, el consumo se cobrará al precio de lista de la API de ese modelo.

Algunos modelos pueden usar [Max Mode](/es/models#max-mode), lo que permite razonamientos más extensos y ventanas de contexto más grandes, de hasta 1M tokens. Aunque la mayoría de las tareas de programación no necesitan usar Max Mode, puede ser útil para consultas más complejas, especialmente con archivos o codebases grandes. Usar Max Mode consumirá más cuota. Puedes ver todas las solicitudes y el desglose de tokens en [tu panel](https://cursor.com/dashboard?tab=usage).

<div id="bugbot">
  ## Bugbot
</div>

Bugbot es un producto independiente de las suscripciones de Cursor y tiene su propio plan de precios.

* **Pro** (40 \$/mes): Revisiones ilimitadas en hasta 200 PR/mes, acceso ilimitado a Cursor Ask, integración con Cursor para corregir bugs y acceso a Bugbot Rules
* **Teams** (40 \$/usuario/mes): Revisiones de código ilimitadas en todos los PR, acceso ilimitado a Cursor Ask, uso compartido entre tu equipo y reglas y configuraciones avanzadas
* **Enterprise** (Personalizado): Todo lo de Teams, más analíticas y reportes avanzados, soporte prioritario y gestión de cuentas

Más información sobre los [precios de Bugbot](https://cursor.com/bugbot#pricing).

<div id="background-agent">
  ## Agente en segundo plano
</div>

Los agentes en segundo plano se cobran según los precios de la API para el [modelo](/es/models) seleccionado. Cuando empieces a usarlos por primera vez, se te pedirá que establezcas un límite de gasto para los agentes en segundo plano.

<Info>
  El cómputo de la máquina virtual (VM) para agentes en segundo plano se tarificará en el futuro.
</Info>

# Admin API
Source: https://docs.cursor.com/es/account/teams/admin-api

Accedé a métricas del equipo, datos de uso e información de gastos vía API

La Admin API te permite acceder de forma programática a los datos de tu equipo, incluyendo información de miembros, métricas de uso y detalles de gastos. Creá dashboards personalizados, herramientas de monitoreo o integrá con tus workflows existentes.

<Note>
  La API está en su primera versión. Estamos ampliando capacidades según el feedback: ¡contanos qué endpoints necesitás!
</Note>

<div id="authentication">
  ## Autenticación
</div>

Todas las solicitudes a la API requieren autenticación con una clave de API. Solo los administradores del equipo pueden crear y gestionar claves de API.

Las claves de API están vinculadas a la organización, son visibles para todos los administradores y no se ven afectadas por el estado de la cuenta de quien las creó originalmente.

<div id="creating-an-api-key">
  ### Crear una clave de API
</div>

1. Ve a **cursor.com/dashboard** → pestaña **Settings** → **Cursor Admin API Keys**
2. Haz clic en **Create New API Key**
3. Dale a tu clave un nombre descriptivo (p. ej., "Integración del panel de uso")
4. Copia la clave generada de inmediato: no la volverás a ver

Formato: `key_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx`

<div id="using-your-api-key">
  ### Usar tu clave de API
</div>

Usa tu clave de API como nombre de usuario en la autenticación básica:

**Usar curl con autenticación básica:**

```bash  theme={null}
curl https://api.cursor.com/{route} -u API_KEY:
```

**O bien establece directamente el encabezado Authorization:**

```bash  theme={null}
Authorization: Basic {base64_encode('API_KEY:')}
```

Todos los endpoints de la API utilizan:

```
https://api.cursor.com
```

<div id="endpoints">
  ## Endpoints
</div>

<div id="get-team-members">
  ### Obtener miembros del equipo
</div>

Obtén todos los miembros del equipo y sus detalles.

```
GET /teams/members
```

#### Respuesta

Devuelve un array de objetos de miembros del equipo:

```typescript  theme={null}
{
  teamMembers: {
    name: string;
    email: string;
    role: 'owner' | 'member' | 'free-owner';
  }[];
}
```

#### Respuesta de ejemplo

```json  theme={null}
{
  "teamMembers": [
    {
      "name": "Alex",
      "email": "developer@company.com",
      "role": "miembro"
    },
    {
      "name": "Sam",
      "email": "admin@company.com",
      "role": "dueño"
    }
  ]
}

```

#### Ejemplo de solicitud

```bash  theme={null}
curl -X GET https://api.cursor.com/teams/members \
  -u TU_CLAVE_API:
```

<div id="get-daily-usage-data">
  ### Obtener datos de uso diario
</div>

Obtén métricas diarias detalladas para tu equipo en un rango de fechas. Ofrece información sobre ediciones de código, uso de la IA de asistencia y tasas de aceptación.

```
POST /teams/uso-diario
```

#### Cuerpo de la solicitud

<Note>
  El rango de fechas no puede superar los 90 días. Haz varias solicitudes para periodos más largos.
</Note>

#### Respuesta

<div id="response-fields">
  #### Campos de respuesta
</div>

#### Ejemplo de respuesta

#### Ejemplo de petición

<div id="get-spending-data">
  ### Obtener datos de gastos
</div>

Obtén la información de gastos del mes calendario actual con búsqueda, ordenación y paginación.

```
POST /teams/spend
```

#### Cuerpo de la solicitud

#### Respuesta

<div id="response-fields">
  #### Campos de respuesta
</div>

#### Ejemplo de respuesta

```json  theme={null}
{
  "teamMemberSpend": [
    {
      "spendCents": 2450,
      "fastPremiumRequests": 1250,
      "name": "Alex",
      "email": "developer@company.com",
      "role": "miembro",
      "hardLimitOverrideDollars": 100
    },
    {
      "spendCents": 1875,
      "fastPremiumRequests": 980,
      "name": "Sam",
      "email": "admin@company.com",
      "role": "propietario",
      "hardLimitOverrideDollars": 0
    },
  ],
  "subscriptionCycleStart": 1708992000000,
  "totalMembers": 15,
  "totalPages": 1
}
```

#### Solicitudes de ejemplo

**Datos básicos de gastos:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/spend \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{}'
```

**Buscar a un usuario específico con paginación:**

<div id="get-usage-events-data">
  ### Obtener datos de eventos de uso
</div>

Obtén eventos de uso detallados para tu equipo con opciones completas de filtrado, búsqueda y paginación. Este endpoint ofrece información granular sobre llamadas individuales a la API, uso de modelos, consumo de tokens y costos.

```
POST /teams/filtered-usage-events
```

#### Cuerpo de la solicitud

#### Respuesta

<div id="response-fields">
  #### Campos de respuesta
</div>

#### Ejemplo de respuesta

```json  theme={null}
{
  "totalUsageEventsCount": 113,
  "pagination": {
    "numPages": 12,
    "currentPage": 1,
    "pageSize": 10,
    "hasNextPage": true,
    "hasPreviousPage": false
  },
  "usageEvents": [
    {
      "timestamp": "1750979225854",
      "model": "claude-4-opus",
      "kind": "Según uso",
      "maxMode": true,
      "requestsCosts": 5,
      "isTokenBasedCall": true,
      "tokenUsage": {
        "inputTokens": 126,
        "outputTokens": 450,
        "cacheWriteTokens": 6112,
        "cacheReadTokens": 11964,
        "totalCents": 20.18232
      },
      "isFreeBugbot": false,
      "userEmail": "developer@company.com"
    },
    {
      "timestamp": "1750979173824",
      "model": "claude-4-opus",
      "kind": "Según uso",
      "maxMode": true,
      "requestsCosts": 10,
      "isTokenBasedCall": true,
      "tokenUsage": {
        "inputTokens": 5805,
        "outputTokens": 311,
        "cacheWriteTokens": 11964,
        "cacheReadTokens": 0,
        "totalCents": 40.16699999999999
      },
      "isFreeBugbot": false,
      "userEmail": "developer@company.com"
    },
    {
      "timestamp": "1750978339901",
      "model": "claude-4-sonnet-thinking",
      "kind": "Incluido en Business"
      "maxMode": true,
      "requestsCosts": 1.4,
      "isTokenBasedCall": false,
      "isFreeBugbot": false,
      "userEmail": "admin@company.com"
    }
  ],
  "period": {
    "startDate": 1748411762359,
    "endDate": 1751003762359
  }
}
```

<div id="example-requests">
  #### Ejemplos de solicitudes
</div>

**Obtener todos los eventos de uso con la paginación por defecto:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/filtered-usage-events \
  -u YOUR_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{}'
```

**Filtrar por intervalo de fechas y usuario específico:**

**Obtén eventos de uso de un usuario específico con paginación personalizada:**

<div id="set-user-spend-limit">
  ### Establecer límite de gasto por usuario
</div>

Establece límites de gasto para miembros específicos del equipo. Esto te permite controlar cuánto puede gastar cada usuario en uso de IA dentro de tu equipo.

```
POST /teams/user-spend-limit
```

<Note>
  **Limitación de frecuencia:** 60 solicitudes por minuto por equipo
</Note>

#### Cuerpo de la solicitud

<Note>
  * El usuario ya debe ser miembro de tu equipo
  * Solo se aceptan valores enteros (sin decimales)
  * Establecer `spendLimitDollars` en 0 fija el límite en \$0
</Note>

#### Respuesta

Devuelve una respuesta estandarizada que indica si hubo éxito o error:

```typescript  theme={null}
{
  outcome: 'success' | 'error';
  message: string;
}
```

<div id="example-responses">
  #### Respuestas de ejemplo
</div>

**Límite establecido correctamente:**

```json  theme={null}
{
  "outcome": "success",
  "message": "Límite de gasto configurado en $100 para el usuario developer@company.com"
}
```

**Respuesta de error:**

```json  theme={null}
{
  "outcome": "error",
  "message": "Formato de correo electrónico inválido"
}
```

<div id="example-requests">
  #### Ejemplos de solicitudes
</div>

**Configurar un límite de gasto:**

```bash  theme={null}
curl -X POST https://api.cursor.com/teams/user-spend-limit \
  -u TU_API_KEY: \
  -H "Content-Type: application/json" \
  -d '{
    "userEmail": "developer@company.com",
    "spendLimitDollars": 100
  }'
```

<div id="repo-blocklists-api">
  ### API de listas de bloqueo de repos
</div>

Agrega repos y usa patrones para evitar que archivos o directorios se indexen o se usen como contexto para tu equipo.

<div id="get-team-repo-blocklists">
  #### Obtener las blocklists de repos del equipo
</div>

Obtén todas las blocklists de repos configuradas para tu equipo.

```
GET /settings/repo-blocklists/repos
```

<div id="response">
  ##### Respuesta
</div>

Devuelve un array de objetos de lista de bloqueo del repositorio:

```typescript  theme={null}
{
  repos: {
    id: string;
    url: string;
    patterns: string[];
  }[];
}
```

<div id="example-response">
  ##### Ejemplo de respuesta
</div>

##### Ejemplo de petición

```bash  theme={null}
curl -X GET https://api.cursor.com/settings/repo-blocklists/repos \
  -u TU_CLAVE_API:
```

<div id="upsert-repo-blocklists">
  #### Upsert de listas de bloqueo de repos
</div>

Reemplaza las listas de bloqueo de repositorios existentes para los repos proporcionados.
*Nota: Este endpoint solo sobrescribirá los patrones de los repositorios proporcionados. Todos los demás repos no se verán afectados.*

```
POST /settings/repo-blocklists/repos/upsert
```

<div id="request-body">
  ##### Cuerpo de la solicitud
</div>

| Parámetro | Tipo  | Requerido | Descripción                                 |
| --------- | ----- | --------- | ------------------------------------------- |
| repos     | array | Sí        | Lista de objetos de bloqueo de repositorios |

Cada objeto de repositorio debe contener:

| Campo    | Tipo      | Requerido | Descripción                                                         |
| -------- | --------- | --------- | ------------------------------------------------------------------- |
| url      | string    | Sí        | URL del repositorio a bloquear                                      |
| patterns | string\[] | Sí        | Lista de patrones de archivos a bloquear (se admiten patrones glob) |

<div id="response">
  ##### Respuesta
</div>

Devuelve la lista actualizada de listas de bloqueo del repositorio:

```typescript  theme={null}
{
  repos: {
    id: string;
    url: string;
    patterns: string[];
  }[];
}
```

<div id="example-request">
  ##### Ejemplo de solicitud
</div>

```bash  theme={null}
curl -X POST https://api.cursor.com/settings/repo-blocklists/repos/upsert \
  -u TU_CLAVE_DE_API: \
  -H "Content-Type: application/json" \
  -d '{
    "repos": [
      {
        "url": "https://github.com/company/sensitive-repo",
        "patterns": ["*.env", "config/*", "secrets/**"]
      },
      {
        "url": "https://github.com/company/internal-tools", 
        "patterns": ["*"]
      }
    ]
  }'
```

<div id="delete-repo-blocklist">
  #### Eliminar la lista de bloqueo de repositorios
</div>

Quita un repositorio específico de la lista de bloqueo.

```
DELETE /settings/repo-blocklists/repos/:repoId
```

<div id="parameters">
  ##### Parámetros
</div>

| Parámetro | Tipo   | Obligatorio | Descripción                                                    |
| --------- | ------ | ----------- | -------------------------------------------------------------- |
| repoId    | string | Sí          | ID de la lista de bloqueo del repositorio que se va a eliminar |

<div id="response">
  ##### Respuesta
</div>

Devuelve 204 No Content si la eliminación se realiza correctamente.

##### Ejemplo de petición

```bash  theme={null}
curl -X DELETE https://api.cursor.com/settings/repo-blocklists/repos/repo_123 \
  -u TU_API_KEY:
```

<div id="pattern-examples">
  #### Ejemplos de patrones
</div>

Patrones comunes de la blocklist:

* `*` - Bloquear todo el repositorio
* `*.env` - Bloquear todos los archivos .env
* `config/*` - Bloquear todos los archivos del directorio config
* `**/*.secret` - Bloquear todos los archivos .secret en cualquier subdirectorio
* `src/api/keys.ts` - Bloquear un archivo específico

# API de seguimiento de código con IA
Source: https://docs.cursor.com/es/account/teams/ai-code-tracking-api

Accede a analíticas de código generadas por IA para los repositorios de tu equipo

Accede a analíticas de código generadas por IA para los repositorios de tu equipo. Incluye uso de IA por commit y cambios de IA aceptados a nivel granular.

<Note>
  La API está en su primera versión. Estamos ampliando funcionalidades según el feedback; ¡cuéntanos qué endpoints necesitas!
</Note>

* **Disponibilidad**: Solo para equipos Enterprise
* **Estado**: Alpha (las estructuras y los campos de la respuesta pueden cambiar)

<div id="authentication">
  ## Autenticación
</div>

Todas las solicitudes a la API requieren autenticación con una clave de API. Esta API usa el mismo método de autenticación del Admin API que otros endpoints.

Para ver instrucciones detalladas de autenticación, consulta [Autenticación del Admin API](/es/account/teams/admin-api#authentication).

Todas las rutas de la API usan:

```
https://api.cursor.com
```

<div id="rate-limits">
  ## Límites de uso
</div>

* 5 solicitudes por minuto por equipo y por endpoint

<div id="query-parameters">
  ## Parámetros de consulta
</div>

Todos los endpoints a continuación aceptan los mismos parámetros a través del query string:

<div className="full-width-table">
  | Parámetro   | Tipo   | Obligatorio | Descripción                                                                                                                                                                            |                                                                                                                              |
  | :---------- | :----- | :---------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------- |
  | `startDate` | string | date        | No                                                                                                                                                                                     | Cadena de fecha ISO, el literal "now" o días relativos como "7d" (equivale a ahora - 7 días). Predeterminado: ahora - 7 días |
  | `endDate`   | string | date        | No                                                                                                                                                                                     | Cadena de fecha ISO, el literal "now" o días relativos como "0d". Predeterminado: ahora                                      |
  | `page`      | number | No          | Número de página (base 1). Predeterminado: 1                                                                                                                                           |                                                                                                                              |
  | `pageSize`  | number | No          | Resultados por página. Predeterminado: 100, máx.: 1000                                                                                                                                 |                                                                                                                              |
  | `user`      | string | No          | Filtro opcional por un único usuario. Acepta email (p. ej., [developer@company.com](mailto:developer@company.com)), ID codificado (p. ej., user\_abc123...) o ID numérico (p. ej., 42) |                                                                                                                              |
</div>

<Note>
  Las respuestas devuelven userId como un ID externo codificado con el prefijo user\_. Esto es estable para el consumo por API.
</Note>

<div id="semantics-and-how-metrics-are-computed">
  ## Semántica y cómo se calculan las métricas
</div>

* **Origen**: "TAB" representa las completions en línea aceptadas; "COMPOSER" representa diffs aceptados desde Composer
* **Métricas de líneas**: tabLinesAdded/Deleted y composerLinesAdded/Deleted se cuentan por separado; nonAiLinesAdded/Deleted se calculan como max(0, totalLines - AI lines)
* **Modo de privacidad**: Si está habilitado en el cliente, puede omitirse cierta metadata (como fileName)
* **Información de la rama**: isPrimaryBranch es true cuando la rama actual es igual a la rama predeterminada del repo; puede ser undefined si la info del repo no está disponible

Podés escanear ese archivo para entender cómo se detectan y reportan los commits y cambios.

<div id="endpoints">
  ## Endpoints
</div>

<div id="get-ai-commit-metrics-json-paginated">
  ### Obtener métricas de commits con IA (JSON, paginadas)
</div>

Obtén métricas agregadas por commit que atribuyen líneas a TAB, COMPOSER y no IA.

```
GET /analytics/ai-code/commits
```

<div id="response">
  #### Respuesta
</div>

```typescript  theme={null}
{
  items: MetricasDeCommitAI[];
  totalCount: number;
  page: number;
  pageSize: number;
}
```

<div id="aicommitmetric-fields">
  #### Campos de AiCommitMetric
</div>

<div className="full-width-table">
  | Campo                  | Tipo    | Descripción                                       |                                          |
  | :--------------------- | :------ | :------------------------------------------------ | ---------------------------------------- |
  | `commitHash`           | string  | Hash del commit de Git                            |                                          |
  | `userId`               | string  | ID de usuario codificado (p. ej., user\_abc123)   |                                          |
  | `userEmail`            | string  | Dirección de correo electrónico del usuario       |                                          |
  | `repoName`             | string  | null                                              | Nombre del repositorio                   |
  | `branchName`           | string  | null                                              | Nombre de la rama                        |
  | `isPrimaryBranch`      | boolean | null                                              | Indica si es la rama principal           |
  | `totalLinesAdded`      | number  | Total de líneas añadidas en el commit             |                                          |
  | `totalLinesDeleted`    | number  | Total de líneas eliminadas en el commit           |                                          |
  | `tabLinesAdded`        | number  | Líneas añadidas mediante autocompletado con TAB   |                                          |
  | `tabLinesDeleted`      | number  | Líneas eliminadas mediante autocompletado con TAB |                                          |
  | `composerLinesAdded`   | number  | Líneas añadidas mediante Composer                 |                                          |
  | `composerLinesDeleted` | number  | Líneas eliminadas mediante Composer               |                                          |
  | `nonAiLinesAdded`      | number  | null                                              | Líneas no generadas por IA añadidas      |
  | `nonAiLinesDeleted`    | number  | null                                              | Líneas no generadas por IA eliminadas    |
  | `message`              | string  | null                                              | Mensaje del commit                       |
  | `commitTs`             | string  | null                                              | Marca de tiempo del commit (formato ISO) |
  | `createdAt`            | string  | Marca de tiempo de ingesta (formato ISO)          |                                          |
</div>

<div id="example-response">
  #### Ejemplo de respuesta
</div>

```json  theme={null}
{
  "items": [
    {
      "commitHash": "a1b2c3d4",
      "userId": "user_3k9x8q...",
      "userEmail": "developer@company.com",
      "repoName": "company/repo",
      "branchName": "main",
      "isPrimaryBranch": true,
      "totalLinesAdded": 120,
      "totalLinesDeleted": 30,
      "tabLinesAdded": 50,
      "tabLinesDeleted": 10,
      "composerLinesAdded": 40,
      "composerLinesDeleted": 5,
      "nonAiLinesAdded": 30,
      "nonAiLinesDeleted": 15,
      "message": "Refactor: extraer cliente de analíticas",
      "commitTs": "2025-07-30T14:12:03.000Z",
      "createdAt": "2025-07-30T14:12:30.000Z"
    }
  ],
  "totalCount": 42,
  "page": 1,
  "pageSize": 100
}
```

<div id="example-requests">
  #### Ejemplos de solicitudes
</div>

**Solicitud básica:**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/commits?startDate=7d&endDate=now&page=1&pageSize=100" \
  -u TU_API_KEY:
```

**Filtrar por usuario (correo):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/commits?startDate=2025-06-01T00:00:00Z&endDate=now&user=developer@company.com" \
  -u TU_API_KEY:
```

<div id="download-ai-commit-metrics-csv-streaming">
  ### Descargar métricas de commits con IA (CSV, streaming)
</div>

Descarga métricas de commits en formato CSV para extracciones de datos de gran volumen.

```
GET /analytics/ai-code/commits.csv
```

<div id="response">
  #### Respuesta
</div>

Encabezados:

* Content-Type: text/csv; charset=utf-8

<div id="csv-columns">
  #### Columnas del CSV
</div>

#### Ejemplo de salida CSV

```csv  theme={null}
commit_hash,user_id,user_email,repo_name,branch_name,is_primary_branch,total_lines_added,total_lines_deleted,tab_lines_added,tab_lines_deleted,composer_lines_added,composer_lines_deleted,non_ai_lines_added,non_ai_lines_deleted,message,commit_ts,created_at
a1b2c3d4,user_3k9x8q...,developer@company.com,company/repo,main,true,120,30,50,10,40,5,30,15,"Refactor: extraer cliente de analítica",2025-07-30T14:12:03.000Z,2025-07-30T14:12:30.000Z
e5f6g7h8,user_3k9x8q...,developer@company.com,company/repo,feature-branch,false,85,15,30,5,25,3,30,7,"Añadir manejo de errores",2025-07-30T13:45:21.000Z,2025-07-30T13:45:45.000Z
```

<div id="example-request">
  #### Ejemplo de solicitud
</div>

```bash  theme={null}
curl -L "https://api.cursor.com/analytics/ai-code/commits.csv?startDate=2025-07-01T00:00:00Z&endDate=now&user=user_3k9x8q..." \
  -u TU_API_KEY: \
  -o commits.csv
```

<div id="get-ai-code-change-metrics-json-paginated">
  ### Obtener métricas de cambios de IA (JSON, paginadas)
</div>

Obtén cambios de IA aceptados a nivel granular, agrupados por un changeId determinista. Útil para analizar eventos de IA aceptados de forma independiente a los commits.

```
GET /analytics/ai-code/changes
```

<div id="response">
  #### Respuesta
</div>

```typescript  theme={null}
{
  items: MetricaCambioCodigoIA[];
  totalCount: number;
  page: number;
  pageSize: number;
}
```

<div id="aicodechangemetric-fields">
  #### Campos de AiCodeChangeMetric
</div>

<div className="full-width-table">
  | Campo               | Tipo   | Descripción                                                              |                                   |
  | :------------------ | :----- | :----------------------------------------------------------------------- | --------------------------------- |
  | `changeId`          | string | ID determinista del cambio                                               |                                   |
  | `userId`            | string | ID de usuario codificado (p. ej., user\_abc123)                          |                                   |
  | `userEmail`         | string | Dirección de correo del usuario                                          |                                   |
  | `source`            | "TAB"  | "COMPOSER"                                                               | Origen del cambio generado por IA |
  | `model`             | string | null                                                                     | Modelo de IA usado                |
  | `totalLinesAdded`   | number | Total de líneas añadidas                                                 |                                   |
  | `totalLinesDeleted` | number | Total de líneas eliminadas                                               |                                   |
  | `createdAt`         | string | Marca de tiempo de ingesta (formato ISO)                                 |                                   |
  | `metadata`          | Array  | Metadatos del archivo (fileName puede omitirse en el modo de privacidad) |                                   |
</div>

#### Respuesta de ejemplo

<div id="example-requests">
  #### Ejemplos de solicitudes
</div>

**Solicitud básica:**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?startDate=14d&endDate=now&page=1&pageSize=200" \
  -u TU_API_KEY:
```

**Filtrar por usuario (ID codificado):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?user=user_3k9x8q..." \
  -u TU_API_KEY:
```

**Filtrar por usuario (correo):**

```bash  theme={null}
curl -X GET "https://api.cursor.com/analytics/ai-code/changes?user=developer@company.com" \
  -u TU_CLAVE_DE_API:
```

<div id="download-ai-code-change-metrics-csv-streaming">
  ### Descargar métricas de cambios de código con IA (CSV, streaming)
</div>

Descarga métricas de cambios en formato CSV para extracciones de datos a gran escala.

```
GET /analytics/ai-code/changes.csv
```

<div id="response">
  #### Respuesta
</div>

Encabezados:

* Content-Type: text/csv; charset=utf-8

<div id="csv-columns">
  #### Columnas del CSV
</div>

<div id="notes">
  #### Notas
</div>

* metadata\_json es un arreglo de entradas de metadatos serializado en JSON (puede omitir fileName en modo de privacidad)
* Al consumir CSV, asegúrate de analizar los campos entre comillas

<div id="sample-csv-output">
  #### Salida de ejemplo en CSV
</div>

<div id="example-request">
  #### Ejemplo de solicitud
</div>

```bash  theme={null}
curl -L "https://api.cursor.com/analytics/ai-code/changes.csv?startDate=30d&endDate=now" \
  -u TU_API_KEY: \
  -o changes.csv
```

<div id="tips">
  ## Consejos
</div>

* Usa el parámetro `user` para filtrar rápido a un solo usuario en todos los endpoints
* Para extracciones de datos grandes, prefiere los endpoints CSV: hacen streaming en páginas de 10.000 registros del lado del servidor
* `isPrimaryBranch` puede ser `undefined` si el cliente no pudo resolver la rama predeterminada
* `commitTs` es la marca de tiempo del commit; `createdAt` es la hora de ingesta en nuestros servidores
* Algunos campos pueden estar ausentes cuando el modo de privacidad está habilitado en el cliente

<div id="changelog">
  ## Registro de cambios
</div>

* **Versión alfa**: Endpoints iniciales para commits y cambios. Las formas de las respuestas pueden evolucionar según el feedback

# Analytics
Source: https://docs.cursor.com/es/account/teams/analytics

Haz seguimiento del uso del equipo y las métrricas de actividad

Los admins del equipo pueden ver métricas desde el [dashboard](/es/account/teams/dashboard).

<div id="total-usage">
  ### Uso total
</div>

Consulta métricas agregadas de tu equipo, incluidas pestañas totales y solicitudes premium. En equipos con menos de 30 días, las métricas muestran el uso desde la creación, incluida la actividad previa al ingreso de los miembros del equipo.

<div id="per-active-user">
  ### Por usuario activo
</div>

Mira los promedios por usuario activo: pestañas aceptadas, líneas de código y solicitudes premium.

<div id="user-activity">
  ### Actividad de usuarios
</div>

Haz seguimiento de usuarios activos semanales y mensuales.

<div id="analytics-report-headers">
  ## Encabezados del informe de analíticas
</div>

Cuando exportas datos de analíticas desde el panel, el informe incluye métricas detalladas sobre el comportamiento de los usuarios y el uso de funcionalidades. Esto es lo que significa cada encabezado:

<div id="user-information">
  ### Información del usuario
</div>

<ResponseField name="Date" type="ISO 8601 timestamp">
  La fecha en que se registraron los datos de analíticas (p. ej., 2024-01-15T04:30:00.000Z)
</ResponseField>

<ResponseField name="User ID" type="string">
  Identificador único de cada usuario en el sistema
</ResponseField>

<ResponseField name="Email" type="string">
  Dirección de correo del usuario asociada a su cuenta
</ResponseField>

<ResponseField name="Is Active" type="boolean">
  Indica si el usuario estuvo activo en esta fecha
</ResponseField>

<div id="ai-generated-code-metrics">
  ### Métricas de código generado por IA
</div>

<ResponseField name="Chat Suggested Lines Added" type="number">
  Total de líneas de código sugeridas por la función de chat con IA
</ResponseField>

<ResponseField name="Chat Suggested Lines Deleted" type="number">
  Total de líneas de código sugeridas para eliminación por el chat con IA
</ResponseField>

<ResponseField name="Chat Accepted Lines Added" type="number">
  Líneas sugeridas por IA que el usuario aceptó y añadió a su código
</ResponseField>

<ResponseField name="Chat Accepted Lines Deleted" type="number">
  Eliminaciones sugeridas por IA que el usuario aceptó
</ResponseField>

<div id="feature-usage-metrics">
  ### Métricas de uso de funcionalidades
</div>

<ResponseField name="Chat Total Applies" type="number">
  Veces que un usuario aplicó cambios generados por IA desde el chat
</ResponseField>

<ResponseField name="Chat Total Accepts" type="number">
  Veces que un usuario aceptó sugerencias de IA
</ResponseField>

<ResponseField name="Chat Total Rejects" type="number">
  Veces que un usuario rechazó sugerencias de IA
</ResponseField>

<ResponseField name="Chat Tabs Shown" type="number">
  Veces que se mostraron al usuario pestañas de sugerencias de IA
</ResponseField>

<ResponseField name="Tabs Accepted" type="number">
  Pestañas de sugerencias de IA aceptadas por el usuario
</ResponseField>

<div id="request-type-metrics">
  ### Métricas por tipo de solicitud
</div>

<ResponseField name="Edit Requests" type="number">
  Solicitudes realizadas a través de la función de composer/edición (Cmd+K ediciones en línea)
</ResponseField>

<ResponseField name="Ask Requests" type="number">
  Solicitudes de chat donde los usuarios hicieron preguntas a la IA
</ResponseField>

<ResponseField name="Agent Requests" type="number">
  Solicitudes realizadas a agentes de IA (asistentes de IA especializados)
</ResponseField>

<ResponseField name="Cmd+K Usages" type="number">
  Veces que se usó la paleta de comandos Cmd+K (o Ctrl+K)
</ResponseField>

<div id="subscription-and-api-metrics">
  ### Métricas de suscripción y API
</div>

<ResponseField name="Subscription Included Reqs" type="number">
  Solicitudes a la IA cubiertas por el plan de suscripción del usuario
</ResponseField>

<ResponseField name="API Key Reqs" type="number">
  Solicitudes realizadas usando claves de API para acceso programático
</ResponseField>

<ResponseField name="Usage-Based Reqs" type="number">
  Solicitudes que computan para la facturación basada en uso
</ResponseField>

<div id="additional-features">
  ### Funcionalidades adicionales
</div>

<ResponseField name="Bugbot Usages" type="number">
  Veces que se usó la función de detección/corrección de errores con IA
</ResponseField>

<div id="configuration-information">
  ### Información de configuración
</div>

<ResponseField name="Most Used Model" type="string">
  El modelo de IA que el usuario usó con mayor frecuencia (p. ej., GPT-4, Claude)
</ResponseField>

<ResponseField name="Most Used Apply Extension" type="string">
  Extensión de archivo más usada al aplicar sugerencias de IA (p. ej., .ts,
  .py, .java)
</ResponseField>

<ResponseField name="Most Used Tab Extension" type="string">
  Extensión de archivo más usada con las funciones de autocompletado por pestañas
</ResponseField>

<ResponseField name="Client Version" type="string">
  Versión del editor Cursor en uso
</ResponseField>

<div id="calculated-metrics">
  ### Métricas calculadas
</div>

El informe también incluye datos procesados que ayudan a entender la contribución de código de la IA:

* Total de líneas agregadas/eliminadas: conteo bruto de todos los cambios de código
* Líneas aceptadas agregadas/eliminadas: líneas que se originaron en sugerencias de IA y fueron aceptadas
* Solicitudes del compositor: solicitudes realizadas a través de la función de compositor en línea
* Solicitudes de chat: solicitudes realizadas a través de la interfaz de chat

<Note>
  Todos los valores numéricos por defecto son 0 si no están presentes, los valores booleanos por defecto son false y los valores de cadena por defecto son cadenas vacías. Las métricas se agregan a nivel diario por usuario.
</Note>

# Analytics V2
Source: https://docs.cursor.com/es/account/teams/analyticsV2

Seguimiento avanzado de métricas de uso y actividad del equipo

Estamos trabajando en el lanzamiento de la V2 de nuestra infraestructura de analytics. Esto incluye un refactor de cómo rastreamos varias métricas.

A partir del **1 de septiembre de 2025**, y para usuarios en **Cursor versión 1.5**, analytics utilizará nuestra infraestructura V2. Las versiones anteriores habrían infracontabilizado varias métricas, incluidas:

* Total de líneas de código aceptadas
* Total de líneas de código sugeridas
* Total de pestañas aceptadas

Mantente al tanto mientras seguimos invirtiendo en analytics y lanzando nuevas funciones en este espacio.

# Panel
Source: https://docs.cursor.com/es/account/teams/dashboard

Administra la facturación, el uso y la configuración del equipo desde tu panel

El panel te permite acceder a la facturación, configurar precios basados en uso y administrar tu equipo.

<div id="overview">
  ## Resumen
</div>

Obtén un resumen rápido de la actividad de tu equipo, las estadísticas de uso y los cambios recientes. La página de resumen ofrece una visión general inmediata de tu espacio de trabajo.

<div id="settings">
  ## Configuración
</div>

Configura las preferencias y la seguridad de todo el equipo. La página de configuración incluye:

<div id="teams-enterprise-settings">
  ## Configuración de Teams y Enterprise
</div>

<AccordionGroup>
  <Accordion title="Privacy Settings">
    Controla las preferencias de uso compartido de datos de tu equipo. Configura políticas de retención cero con proveedores de IA (OpenAI, Anthropic, Google Vertex AI, xAI Grok) y gestiona la aplicación de la privacidad a nivel de equipo.
  </Accordion>

{" "}

<Accordion title="Usage-Based Pricing Settings">
    Activa la facturación basada en uso y establece límites de gasto. Configura límites de gasto mensuales del equipo y límites opcionales por usuario. Controla si solo los admins pueden modificar esta configuración.
  </Accordion>

{" "}

<Accordion title="Bedrock IAM Role">
    Configura roles de AWS Bedrock IAM para una integración segura en la nube.
  </Accordion>

{" "}

<Accordion title="Single Sign-On (SSO)">
    Configura la autenticación SSO para equipos Enterprise y agiliza el acceso de los usuarios y la seguridad.
  </Accordion>

{" "}

<Accordion title="Cursor Admin API Keys">
    Crea y gestiona claves de API para el acceso programático a las funciones de administración de Cursor.
  </Accordion>

{" "}

<Accordion title="Active Sessions">
    Supervisa y gestiona las sesiones activas de usuarios en todo tu equipo.
  </Accordion>

<Accordion title="Invite Code Management">
    Crea y gestiona códigos de invitación para añadir nuevas personas al equipo.
  </Accordion>

<Accordion title="API Endpoints">
    Accede a los endpoints de la REST API de Cursor para la integración programática. Todos los endpoints de la API están disponibles en los planes Team y Enterprise, excepto la [AI Code Tracking API](/es/docs/account/teams/ai-code-tracking-api), que requiere una suscripción Enterprise.
  </Accordion>
</AccordionGroup>

<div id="enterprise-only-settings">
  ## Configuraciones exclusivas de Enterprise
</div>

<AccordionGroup>
  {" "}

<Accordion title="Model Access Control">
    Controla qué modelos de IA están disponibles para los miembros del equipo. Establece restricciones en
    modelos específicos o niveles de modelo para gestionar costos y garantizar un uso adecuado
    en toda tu organización.
  </Accordion>

{" "}

<Accordion title="Auto Run Configuration (0.49+)">
    Configura los ajustes de ejecución automática de comandos para Cursor versión 0.49 y
    posteriores. Controla qué comandos pueden ejecutarse automáticamente y define políticas de seguridad
    para la ejecución de código.
  </Accordion>

<Accordion title="Repository Blocklist">
    Evita el acceso a repositorios específicos por razones de seguridad o cumplimiento.
  </Accordion>

{" "}

<Accordion title="MCP Configuration (0.51+)">
    Configura los ajustes de Model Context Protocol para Cursor versión 0.51 y posteriores.
    Administra cómo los modelos acceden y procesan el contexto desde tu entorno de
    desarrollo.
  </Accordion>

{" "}

<Accordion title="Cursor Ignore Configuration (0.50+)">
    Configura patrones de exclusión para archivos y directorios en Cursor versión 0.50 y
    posteriores. Controla qué archivos y directorios se excluyen del análisis de IA y de
    las sugerencias.
  </Accordion>

<Accordion title=".cursor Directory Protection (0.51+)">
    Protege el directorio .cursor contra accesos no autorizados en la versión 0.51 y posteriores. Asegura que los archivos sensibles de configuración y caché permanezcan protegidos.
  </Accordion>

<Accordion title="AI Code Tracking API">
    Accede a analíticas detalladas del código generado por IA para los repositorios de tu equipo. Obtén métricas de uso de IA por commit y cambios de IA aceptados a nivel granular mediante endpoints de la API REST. Requiere un plan Enterprise. Encuentra más información [aquí](/es/account/teams/ai-code-tracking-api).
  </Accordion>
</AccordionGroup>

<Note>
  El aprovisionamiento **SCIM** (System for Cross-domain Identity Management) también está
  disponible para planes Enterprise. Consulta nuestra [documentación de SCIM](/es/account/teams/scim)
  para obtener instrucciones de configuración.
</Note>

<div id="members">
  ## Miembros
</div>

Administra a los miembros de tu equipo, invita a nuevos usuarios y controla los permisos de acceso. Configura permisos por rol y monitorea la actividad de los miembros.

<div id="integrations">
  ## Integraciones
</div>

{" "}

Conecta Cursor con tus herramientas y servicios favoritos. Configura integraciones con sistemas de control de versiones, herramientas de gestión de proyectos y otros servicios para desarrolladores.

<div id="background-agents">
  ## Agentes en segundo plano
</div>

{" "}

Monitorea y gestiona los agentes en segundo plano que se ejecutan en tu espacio de trabajo. Revisa el estado de los agentes, los registros y el uso de recursos.

<div id="bugbot">
  ## Bugbot
</div>

Accedé a funciones automatizadas para detectar y corregir bugs. Bugbot te ayuda a identificar y resolver problemas comunes en tu codebase automáticamente.

<div id="active-directory-management">
  ## Gestión de Active Directory
</div>

Para equipos empresariales, gestiona la autenticación de usuarios y el acceso mediante la integración con Active Directory. Configura SSO y el aprovisionamiento de usuarios.

Haz un seguimiento detallado de métricas de uso, como solicitudes a la IA, uso de modelos y consumo de recursos. Supervisa el uso entre los miembros del equipo y los proyectos.

<div id="billing-invoices">
  ## Facturación e facturas
</div>

Gestiona tu suscripción, actualiza los métodos de pago y accede al historial de facturación. Descarga las facturas y administra la configuración de precios según el uso.

# Ajustes de empresa
Source: https://docs.cursor.com/es/account/teams/enterprise-settings

Administra centralmente los ajustes de Cursor para tu organización

<div id="enterprise-settings">
  # Configuración para empresas
</div>

Podés gestionar de forma centralizada funciones específicas de Cursor mediante soluciones de administración de dispositivos para asegurarte de que cumpla con las necesidades de tu organización. Cuando especificás una política de Cursor, su valor reemplaza la configuración correspondiente de Cursor en los dispositivos de los usuarios.

Editor de configuración que muestra que la opción 'Extensions: Allowed' está gestionada por la organización.

Actualmente, Cursor ofrece políticas para controlar las siguientes funciones administradas por el equipo de TI:

| Policy            | Description                                                                                                                         | Cursor setting           | Available since |
| ----------------- | ----------------------------------------------------------------------------------------------------------------------------------- | ------------------------ | --------------- |
| AllowedExtensions | Controla qué extensiones se pueden instalar.                                                                                        | extensions.allowed       | 1.2             |
| AllowedTeamId     | Controla qué IDs de equipo pueden iniciar sesión. Los usuarios con IDs de equipo no autorizados se cierran sesión de forma forzosa. | cursorAuth.allowedTeamId | 1.3             |

<div id="configure-allowed-extensions">
  ## Configurar extensiones permitidas
</div>

La configuración de Cursor `extensions.allowed` controla qué extensiones se pueden instalar. Esta configuración acepta un objeto JSON donde las claves son nombres de publishers y los valores son booleanos que indican si se permiten las extensiones de ese publisher.

Por ejemplo, establecer `extensions.allowed` en `{"anysphere": true, "github": true}` permite extensiones de los publishers Anysphere y GitHub, mientras que configurarlo en `{"anysphere": false}` bloquea las extensiones de Anysphere.

Para gestionar de forma centralizada las extensiones permitidas para tu organización, configura la política `AllowedExtensions` usando tu solución de administración de dispositivos. Esta política reemplaza la configuración `extensions.allowed` en los dispositivos de los usuarios. El valor de esta política es una cadena JSON que define los publishers permitidos.

Si quieres saber más sobre las extensiones en Cursor, consulta la documentación de extensiones.

<div id="configure-allowed-team-ids">
  ## Configurar IDs de equipo permitidos
</div>

La configuración de Cursor `cursorAuth.allowedTeamId` controla qué IDs de equipo pueden iniciar sesión en Cursor. Esta configuración acepta una lista de IDs de equipo autorizados separada por comas.

Por ejemplo, establecer `cursorAuth.allowedTeamId` en "1,3,7" permite que usuarios de esos IDs de equipo específicos inicien sesión.

Cuando un usuario intenta iniciar sesión con un ID de equipo que no está en la lista permitida:

* Se cierra su sesión de forma inmediata
* Se muestra un mensaje de error
* La aplicación bloquea más intentos de autenticación hasta que se use un ID de equipo válido

Para gestionar de forma centralizada los IDs de equipo permitidos de tu organización, configura la directiva `AllowedTeamId` usando tu solución de gestión de dispositivos. Esta directiva reemplaza la configuración `cursorAuth.allowedTeamId` en los dispositivos de los usuarios. El valor de esta directiva es una cadena que contiene la lista de IDs de equipo autorizados separada por comas.

<div id="group-policy-on-windows">
  ## Directiva de grupo en Windows
</div>

Cursor es compatible con la Directiva de grupo basada en el Registro de Windows. Cuando se instalan las definiciones de directiva, los administradores pueden usar el Editor de directivas de grupo local para gestionar los valores de las directivas.

Para agregar una directiva:

1. Copia los archivos ADMX y ADML de `AppData\Local\Programs\cursor\policies`.
2. Pega el archivo ADMX en `C:\Windows\PolicyDefinitions` y el archivo ADML en `C:\Windows\PolicyDefinitions\<your-locale>\`.
3. Reinicia el Editor de directivas de grupo local.
4. Establece los valores de directiva correspondientes (p. ej., `{"anysphere": true, "github": true}` para la directiva `AllowedExtensions`) en el Editor de directivas de grupo local.

Las directivas se pueden establecer tanto a nivel de equipo como a nivel de usuario. Si ambos están configurados, el nivel de equipo tendrá prioridad. Cuando se establece un valor de directiva, ese valor reemplaza el valor de configuración de Cursor establecido en cualquier nivel (predeterminado, usuario, espacio de trabajo, etc.).

<div id="configuration-profiles-on-macos">
  ## Perfiles de configuración en macOS
</div>

Los perfiles de configuración administran ajustes en dispositivos macOS. Un perfil es un archivo XML con pares clave/valor que corresponden a las políticas disponibles. Estos perfiles pueden desplegarse con soluciones de Mobile Device Management (MDM) o instalarse manualmente.

<Accordion title="Archivo .mobileconfig de ejemplo">
  A continuación se muestra un archivo `.mobileconfig` de ejemplo para macOS:

<div id="string-policies">
  ### Políticas de tipo string
</div>

El ejemplo de abajo muestra cómo configurar la política `AllowedExtensions`. El valor de la política comienza vacío en el archivo de ejemplo (no se permite ninguna extensión).

```
<key>AllowedExtensions</key>
<string></string>
```

Agrega la cadena JSON adecuada que defina tu policy entre las etiquetas `<string>`.

```
<key>AllowedExtensions</key>
<string>{"anysphere": true, "github": true}</string>
```

Para la directiva `AllowedTeamId`, agrega la lista de IDs de equipo separadas por comas:

```
<key>AllowedTeamId</key>
<string>1,3,7</string>
```

**Importante:** El archivo `.mobileconfig` proporcionado inicializa **todas** las políticas disponibles en esa versión de Cursor. Elimina cualquier política que no necesites.

Si no editas o eliminas una política del `.mobileconfig` de muestra, esa política se aplicará con su valor predeterminado (restrictivo).

Instala manualmente un perfil de configuración haciendo doble clic en el perfil `.mobileconfig` en Finder y luego habilitándolo en Preferencias del Sistema en **General** > **Administración de dispositivos**. Al quitar el perfil desde Preferencias del Sistema, se eliminarán las políticas de Cursor.

Para obtener más información sobre los perfiles de configuración, consulta la documentación de Apple.

<div id="additional-policies">
  ## Políticas adicionales
</div>

La idea es promover las configuraciones actuales de Cursor como políticas y ceñirse de cerca a las existentes, para que la nomenclatura y el comportamiento sean consistentes. Si querés proponer más políticas, abrí un issue en el repositorio de GitHub de Cursor. El equipo va a determinar si ya existe una configuración correspondiente para ese comportamiento o si se debería crear una nueva para controlar el comportamiento deseado.

<div id="frequently-asked-questions">
  ## Preguntas frecuentes
</div>

<div id="does-cursor-support-configuration-profiles-on-linux">
  ### ¿Cursor admite perfiles de configuración en Linux?
</div>

El soporte para Linux no está en la hoja de ruta. Si te interesan los perfiles de configuración en Linux, abre un issue en el repositorio de GitHub de Cursor y comparte detalles sobre tu caso.

# Miembros y roles
Source: https://docs.cursor.com/es/account/teams/members

Administra los miembros y roles del equipo

Los equipos de Cursor tienen tres roles:

<div id="roles">
  ## Roles
</div>

**Miembros** es el rol predeterminado con acceso a las funciones Pro de Cursor.

* Acceso completo a las funciones Pro de Cursor
* Sin acceso a la configuración de facturación ni al panel de administración
* Pueden ver su propio uso y el presupuesto restante según el uso

**Admins** controlan la gestión del equipo y la configuración de seguridad.

* Acceso completo a las funciones Pro
* Agregar/quitar miembros, modificar roles, configurar SSO
* Configurar precios basados en uso y límites de gasto
* Acceso a analíticas del equipo

**Admins no pagados** gestionan equipos sin ocupar un asiento de pago; ideal para personal de TI o finanzas que no necesita acceso a Cursor.

* No facturables, sin funciones Pro
* Las mismas capacidades administrativas que los Admins

<Info>Los Admins no pagados requieren al menos un usuario de pago en el equipo.</Info>

<div id="role-comparison">
  ## Comparación de roles
</div>

<div className="full-width-table">
  | Capacidades                 | Miembro | Admin | Admin sin pago |
  | --------------------------- | :-----: | :---: | :------------: |
  | Usar funciones de Cursor    |    ✓    |   ✓   |                |
  | Invitar miembros            |    ✓    |   ✓   |        ✓       |
  | Quitar miembros             |         |   ✓   |        ✓       |
  | Cambiar rol de usuario      |         |   ✓   |        ✓       |
  | Panel de administración     |         |   ✓   |        ✓       |
  | Configurar SSO/seguridad    |         |   ✓   |        ✓       |
  | Gestionar facturación       |         |   ✓   |        ✓       |
  | Ver analíticas              |         |   ✓   |        ✓       |
  | Gestionar acceso            |         |   ✓   |        ✓       |
  | Establecer controles de uso |         |   ✓   |        ✓       |
  | Requiere asiento de pago    |    ✓    |   ✓   |                |
</div>

<div id="managing-members">
  ## Gestión de miembros
</div>

Todos los miembros del equipo pueden invitar a otras personas. Actualmente no controlamos las invitaciones.

<div id="add-member">
  ### Agregar miembro
</div>

Podés agregar miembros de tres maneras:

1. **Invitación por email**

* Hacé clic en `Invite Members`
   * Ingresá direcciones de email
   * Les usuaries reciben invitaciones por email

2. **Enlace de invitación**

* Hacé clic en `Invite Members`
   * Copiá `Invite Link`
   * Compartilo con miembros del equipo

3. **SSO**
   * Configurá SSO en el [admin dashboard](/es/account/teams/sso)
   * Les usuaries se unen automáticamente al iniciar sesión con el email de SSO

<Warning>
  Los enlaces de invitación tienen una fecha de vencimiento larga: cualquiera con el enlace puede unirse.
  Revocalos o usá [SSO](/es/account/teams/sso)
</Warning>

<div id="remove-member">
  ### Eliminar miembro
</div>

Les admins pueden eliminar miembros en cualquier momento mediante el menú contextual → "Remove". Si un miembro usó créditos, su asiento permanece ocupado hasta el final del ciclo de facturación.

<div id="change-role">
  ### Cambiar rol
</div>

Les admins pueden cambiar los roles de otres miembros haciendo clic en el menú contextual y luego usando la opción "Change role".<br />

Debe haber al menos un admin y un miembro de pago en el equipo en todo momento.

<div id="security-sso">
  ## Seguridad y SSO
</div>

El inicio de sesión único (SSO) con SAML 2.0 está disponible en los planes Team. Las funciones clave incluyen:

* Configurar conexiones de SSO ([más info](/es/account/teams/sso))
* Configurar la verificación de dominio
* Alta automática de usuarios
* Opciones para exigir SSO
* Integración con el proveedor de identidad (Okta, etc.)

<Note>
  <p className="!mb-0">La verificación de dominio es necesaria para habilitar SSO.</p>
</Note>

<div id="usage-controls">
  ## Controles de uso
</div>

Accede a la configuración de uso para:

* Habilitar la facturación basada en el uso
* Habilitar para modelos premium
* Permitir modificaciones solo para administradores
* Establecer límites de gasto mensuales
* Supervisar el uso de todo el equipo

<div id="billing">
  ## Facturación
</div>

Al agregar miembros al equipo:

* Cada miembro o admin suma un asiento facturable (ver [pricing](https://cursor.com/pricing))
* A los nuevos miembros se les cobra de forma prorrateada por el tiempo restante del período de facturación
* Los asientos de admin no pago no se cuentan

Las altas a mitad de mes solo cobran por los días usados. Al quitar miembros que ya usaron créditos, su asiento queda ocupado hasta el final del ciclo de facturación: no hay reembolsos prorrateados.

Los cambios de rol (p. ej., de Admin a Admin no pago) ajustan la facturación desde la fecha del cambio. Elegí facturación mensual o anual.

La renovación mensual/anual ocurre en tu fecha original de registro, sin importar los cambios de miembros.

<div id="switch-to-yearly-billing">
  ### Cambiar a facturación anual
</div>

Ahorrá un **20%** al pasar de mensual a anual:

1. Andá al [Dashboard](https://cursor.com/dashboard)
2. En la sección de cuenta, hacé clic en "Advanced" y luego en "Upgrade to yearly billing"

<Note>
  Solo podés cambiar de mensual a anual desde el dashboard. Para cambiar de
  anual a mensual, escribí a [hi@cursor.com](mailto:hi@cursor.com).
</Note>

# SCIM
Source: https://docs.cursor.com/es/account/teams/scim

Configurá el aprovisionamiento de SCIM para la gestión automatizada de usuarios y grupos

<div id="overview">
  ## Descripción general
</div>

El aprovisionamiento con SCIM 2.0 gestiona automáticamente los miembros de tu equipo y los grupos del directorio a través de tu proveedor de identidad. Disponible en planes Enterprise con SSO activado.

<product_visual type="screenshot">
  Panel de SCIM que muestra la configuración de Active Directory Management
</product_visual>

<div id="prerequisites">
  ## Requisitos previos
</div>

* Plan de Cursor Enterprise
* Primero debe configurarse el SSO: **SCIM requiere una conexión SSO activa**
* Acceso de administrador a tu proveedor de identidad (Okta, Azure AD, etc.)
* Acceso de administrador a tu organización de Cursor

<div id="how-it-works">
  ## Cómo funciona
</div>

<div id="user-provisioning">
  ### Aprovisionamiento de usuarios
</div>

Los usuarios se añaden automáticamente a Cursor cuando los asignas a la aplicación SCIM en tu proveedor de identidad. Al desasignarlos, se eliminan. Los cambios se sincronizan en tiempo real.

<div id="directory-groups">
  ### Grupos de directorio
</div>

Los grupos de directorio y su pertenencia se sincronizan desde tu proveedor de identidad. La gestión de grupos y usuarios debe hacerse a través de tu proveedor de identidad; Cursor muestra esta información como de solo lectura.

<div id="spend-management">
  ### Gestión del gasto
</div>

Configura distintos límites de gasto por usuario para cada grupo de directorio. Los límites de los grupos de directorio tienen prioridad sobre los límites a nivel de equipo. Los usuarios que pertenezcan a varios grupos reciben el límite de gasto más alto aplicable.

<div id="setup">
  ## Configuración
</div>

<Steps>
  <Step title="Asegurate de tener SSO configurado">
    SCIM requiere que primero configures SSO. Si aún no lo hiciste,
    seguí la [guía de configuración de SSO](/es/account/teams/sso) antes de continuar.
  </Step>

<Step title="Accedé a Active Directory Management">
    Navegá a
    [cursor.com/dashboard?tab=active-directory](https://www.cursor.com/dashboard?tab=active-directory)
    con una cuenta de admin, o andá a la configuración de tu dashboard y seleccioná la
    "Active Directory Management" tab.
  </Step>

<Step title="Iniciá la configuración de SCIM">
    Una vez verificado SSO, vas a ver un enlace para la configuración de SCIM paso a paso. Hacé clic
    para iniciar el asistente de configuración.
  </Step>

<Step title="Configurá SCIM en tu proveedor de identidad">
    En tu proveedor de identidad: Creá o configurá tu aplicación SCIM. Usá
    el endpoint y el token de SCIM provistos por Cursor. Activá el aprovisionamiento de usuarios y el envío de grupos.
    Probá la conexión.
  </Step>

<Step title="Configurá límites de gasto (opcional)">
    De vuelta en la página de Active Directory Management de Cursor: Mirá tus grupos de directorio sincronizados.
    Configurá límites de gasto por usuario para grupos específicos según sea necesario.
    Revisá qué límites aplican a usuarios en múltiples grupos.
  </Step>
</Steps>

<div id="identity-provider-setup">
  ### Configuración del proveedor de identidad
</div>

Para instrucciones específicas del proveedor:

<Card title="Guías del proveedor de identidad" icon="book" href="https://workos.com/docs/integrations">
  Instrucciones de configuración para Okta, Azure AD, Google Workspace y más.
</Card>

<div id="managing-users-and-groups">
  ## Gestión de usuarios y grupos
</div>

<Warning>
  Toda la gestión de usuarios y grupos debe hacerse a través de tu proveedor de identidad.
  Los cambios que realices en tu proveedor de identidad se sincronizan automáticamente con Cursor, pero
  no puedes modificar usuarios ni grupos directamente en Cursor.
</Warning>

<div id="user-management">
  ### Gestión de usuarios
</div>

* Agrega usuarios asignándolos a tu aplicación SCIM en tu proveedor de identidad
* Elimina usuarios quitándolos de la asignación de la aplicación SCIM
* Los cambios en el perfil del usuario (nombre, correo) se sincronizan automáticamente desde tu proveedor de identidad

<div id="group-management">
  ### Gestión de grupos
</div>

* Los grupos del directorio se sincronizan automáticamente desde tu proveedor de identidad
* Los cambios en la pertenencia a grupos se reflejan en tiempo real
* Usa grupos para organizar usuarios y establecer distintos límites de gasto

<div id="spend-limits">
  ### Límites de gasto
</div>

* Establece distintos límites por usuario para cada grupo del directorio
* Los usuarios heredan el límite de gasto más alto de sus grupos
* Los límites de grupo reemplazan el límite predeterminado por usuario del equipo

<div id="faq">
  ## Preguntas frecuentes
</div>

<div id="why-isnt-scim-management-showing-up-in-my-dashboard">
  ### ¿Por qué la administración de SCIM no aparece en mi panel?
</div>

Asegurate de que SSO esté configurado correctamente y funcionando antes de configurar SCIM. SCIM requiere una conexión de SSO activa para funcionar.

<div id="why-arent-users-syncing">
  ### ¿Por qué los usuarios no se están sincronizando?
</div>

Verificá que los usuarios estén asignados a la aplicación de SCIM en tu proveedor de identidad. Los usuarios tienen que estar asignados explícitamente para aparecer en Cursor.

<div id="why-arent-groups-appearing">
  ### ¿Por qué los grupos no aparecen?
</div>

Comprobá que el aprovisionamiento de grupos por push esté habilitado en la configuración de SCIM de tu proveedor de identidad. La sincronización de grupos se configura por separado de la sincronización de usuarios.

<div id="why-arent-spend-limits-applying">
  ### ¿Por qué no se aplican los límites de gasto?
</div>

Confirmá que los usuarios estén correctamente asignados a los grupos correspondientes en tu proveedor de identidad. La pertenencia a grupos determina qué límites de gasto se aplican.

<div id="can-i-manage-scim-users-and-groups-directly-in-cursor">
  ### ¿Puedo administrar usuarios y grupos de SCIM directamente en Cursor?
</div>

No. Toda la administración de usuarios y grupos se tiene que hacer a través de tu proveedor de identidad. Cursor muestra esta información como de solo lectura.

<div id="how-quickly-do-changes-sync">
  ### ¿Qué tan rápido se sincronizan los cambios?
</div>

Los cambios realizados en tu proveedor de identidad se sincronizan con Cursor en tiempo real. Puede haber una breve demora en operaciones masivas.

# Primeros pasos
Source: https://docs.cursor.com/es/account/teams/setup

Crea y configura un equipo de Cursor

<div id="cursor-for-teams">
  ## Cursor para equipos
</div>

Cursor funciona para personas y equipos. El plan Teams ofrece herramientas para organizaciones: SSO, gestión de equipos, controles de acceso y análisis de uso.

<div id="creating-a-team">
  ## Crear un equipo
</div>

Crea un equipo siguiendo estos pasos:

<Steps>
  <Step title="Configurar el plan Teams">
    Para crear un equipo, sigue estos pasos:

1. **Para usuarios nuevos**: Visita [cursor.com/team/new-team](https://cursor.com/team/new-team) para crear una cuenta y un equipo nuevos
    2. **Para usuarios existentes**: Ve a tu [panel](/es/account/dashboard) y haz clic en "Upgrade to Teams"
  </Step>

<Step title="Ingresar detalles del equipo">
    Elige un nombre para el equipo y el ciclo de facturación

<Step title="Invitar miembros">
    Invita a miembros del equipo. Los usuarios se prorratean: solo pagas por el tiempo que son miembros.

<Step title="Habilitar SSO (opcional)">
    Habilita el [SSO](/es/account/teams/sso) para mayor seguridad y onboarding automatizado.

<div id="faq">
  ## Preguntas frecuentes
</div>

<AccordionGroup>
  <Accordion title="Mi equipo usa Zscaler / un proxy / una VPN, ¿funcionará Cursor?">
    Cursor usa HTTP/2 de forma predeterminada. Algunos proxies y VPN lo bloquean.

Activa la opción de usar HTTP/1.1 como respaldo en la configuración.
  </Accordion>

<Accordion title="¿Cómo puedo comprar licencias para mi empresa?">
    Cursor factura por usuario activo, no por plazas. Agrega o quita usuarios en cualquier momento: a los nuevos miembros se les cobra de forma prorrateada por el tiempo restante. Si un usuario eliminado ha usado créditos, su plaza sigue ocupada hasta el final del ciclo de facturación.

Tu fecha de renovación no cambia.
  </Accordion>

<Accordion title="¿Cómo puedo configurar un equipo si no estoy usando Cursor?">
    Ponte como [Unpaid Admin](/es/account/teams/members) para administrar sin una licencia.

<Warning>
      Los equipos necesitan al menos un miembro de pago. Puedes configurar, invitar a un miembro y luego cambiar tu rol antes de la facturación.
    </Warning>
  </Accordion>

<Accordion title="¿Cómo puedo agregar Cursor al MDM de mi empresa?">
    Los enlaces de descarga para todas las plataformas están disponibles en [cursor.com/downloads](https://cursor.com/downloads).

Instrucciones de MDM:

* [Omnissa Workspace ONE](https://docs.omnissa.com/bundle/MobileApplicationManagementVSaaS/page/DeployInternalApplications.html) (antes VMware)
    * [Microsoft Intune (Windows)](https://learn.microsoft.com/en-us/mem/intune-service/apps/apps-win32-app-management)
    * [Microsoft Intune (Mac)](https://learn.microsoft.com/en-us/mem/intune-service/apps/lob-apps-macos-dmg)
    * [Kandji MDM](https://support.kandji.io/kb/custom-apps-overview)
  </Accordion>
</AccordionGroup>

# SSO
Source: https://docs.cursor.com/es/account/teams/sso

Configura el inicio de sesión único para tu equipo

<div id="overview">
  ## Descripción general
</div>

El SSO con SAML 2.0 está disponible sin costo adicional en los planes Business. Usa tu proveedor de identidad (IdP) existente para autenticar a los miembros del equipo sin que necesiten cuentas independientes de Cursor.

<div id="prerequisites">
  ## Requisitos previos
</div>

* Plan Cursor Team
* Acceso de admin a tu proveedor de identidad (p. ej., Okta)
* Acceso de admin a tu organización de Cursor

<div id="configuration-steps">
  ## Pasos de configuración
</div>

<Steps>
  <Step title="Inicia sesión en tu cuenta de Cursor">
    Ve a [cursor.com/dashboard?tab=settings](https://www.cursor.com/dashboard?tab=settings) con una cuenta de admin.
  </Step>

<Step title="Ubica la configuración de SSO">
    Busca la sección "Single Sign-On (SSO)" y ábrela.
  </Step>

<Step title="Comienza el proceso de configuración">
    Haz clic en el botón "SSO Provider Connection settings" para iniciar la configuración de SSO y sigue el asistente.
  </Step>

<Step title="Configura tu proveedor de identidad">
    En tu proveedor de identidad (p. ej., Okta):

* Crea una nueva aplicación SAML
    * Configura los parámetros de SAML usando la información de Cursor
    * Configura el aprovisionamiento Just‑in‑Time (JIT)
  </Step>

<Step title="Verifica el dominio">
    Verifica el dominio de tus usuarios en Cursor haciendo clic en el botón "Domain verification settings".
  </Step>
</Steps>

<div id="identity-provider-setup-guides">
  ### Guías de configuración del proveedor de identidad
</div>

Para instrucciones de configuración específicas por proveedor:

<Card title="Guías del proveedor de identidad" icon="book" href="https://workos.com/docs/integrations">
  Instrucciones de configuración para Okta, Azure AD, Google Workspace y más.
</Card>

<div id="additional-settings">
  ## Configuraciones adicionales
</div>

* Controla la aplicación forzosa de SSO desde el panel de administración
* Los nuevos usuarios se registran automáticamente al iniciar sesión con SSO
* Administra a los usuarios a través de tu proveedor de identidad

<div id="troubleshooting">
  ## Solución de problemas
</div>

Si tienes problemas:

* Verifica que el dominio esté validado en Cursor
* Asegúrate de que los atributos SAML estén asignados correctamente
* Comprueba que el SSO esté habilitado en el panel de administración
* Asegúrate de que nombre y apellido coincidan entre el proveedor de identidad y Cursor
* Revisa las guías específicas del proveedor de más arriba
* Escríbenos a [hi@cursor.com](mailto:hi@cursor.com) si el problema persiste

# Acceso a actualizaciones
Source: https://docs.cursor.com/es/account/update-access

Elige con qué frecuencia quieres recibir actualizaciones

Cursor tiene dos canales de actualización.

<Tabs>
  <Tab title="Default">
    El canal de actualización predeterminado con versiones comprobadas.

* Versiones estables
    * Correcciones de errores de las pruebas previas al lanzamiento
    * Predeterminado para todos los usuarios
    * Única opción para usuarios de equipo

<Note>
      Las cuentas de equipo y empresariales usan el modo Default.
    </Note>
  </Tab>

<Tab title="Early Access">
    Versiones preliminares con nuevas funciones.

<Warning>
      Las versiones de Early Access pueden tener errores o problemas de estabilidad.
    </Warning>

* Acceso a funciones en desarrollo
    * Puede contener errores
    * No disponible para cuentas de equipo
  </Tab>
</Tabs>

<div id="change-update-channel">
  ## Cambiar el canal de actualización
</div>

1. **Abrir configuración**: Presiona <Kbd>Cmd+Shift+J</Kbd>
2. **Ir a Beta**: Selecciona Beta en la barra lateral
3. **Seleccionar canal**: Elige Default o Early Access

Reporta problemas de Early Access en el [Forum](https://forum.cursor.com).

# Apply
Source: https://docs.cursor.com/es/agent/apply

Aprende a aplicar, aceptar o rechazar sugerencias de código desde el chat con Apply

<div id="how-apply-works">
  ## Cómo funciona Apply
</div>

Apply es un modelo especializado de Cursor que toma el código generado en el chat y lo integra en tus archivos. Procesa los bloques de código de las conversaciones y aplica los cambios a tu código.

Apply no genera código por sí mismo. El modelo de chat genera el código y Apply se encarga de integrarlo en los archivos existentes. Puede procesar cambios en múltiples archivos y en bases de código grandes.

<div id="apply-code-blocks">
  ## Aplicar bloques de código
</div>

Para aplicar una sugerencia de bloque de código, haz clic en el botón de reproducir en la esquina superior derecha del bloque.

# Checkpoints
Source: https://docs.cursor.com/es/agent/chat/checkpoints

Guarda y restaura estados anteriores después de cambios del Agent

Los checkpoints son instantáneas automáticas de los cambios que el Agent hace en tu base de código. Te permiten deshacer las modificaciones del Agent cuando lo necesites.

<div id="restoring-checkpoints">
  ## Restaurar checkpoints
</div>

Dos formas de restaurar:

1. **Desde el cuadro de entrada**: haz clic en el botón `Restore Checkpoint` en solicitudes anteriores
2. **Desde el mensaje**: haz clic en el botón + al pasar el cursor sobre un mensaje

<Warning>
  Los checkpoints no son control de versiones. Usa Git para el historial permanente.
</Warning>

<div id="how-they-work">
  ## Cómo funcionan
</div>

* Se guardan localmente, aparte de Git
* Solo registran cambios del Agent (no ediciones manuales)
* Se limpian automáticamente

<Note>
  Las ediciones manuales no se registran. Usa checkpoints solo para cambios del Agent.
</Note>

<div id="faq">
  ## Preguntas frecuentes
</div>

<AccordionGroup>
  <Accordion title="¿Los checkpoints afectan a Git?">
    No. Son independientes del historial de Git.
  </Accordion>

{" "}

<Accordion title="¿Cuánto tiempo se conservan?">
    Solo durante la sesión actual y el historial reciente. Se eliminan automáticamente.
  </Accordion>

<Accordion title="¿Puedo crearlos manualmente?">
    No. Cursor los crea automáticamente.
  </Accordion>
</AccordionGroup>

{" "}

# Comandos
Source: https://docs.cursor.com/es/agent/chat/commands

Define comandos para flujos de trabajo reutilizables

Los comandos personalizados te permiten crear flujos de trabajo reutilizables que se pueden activar con un simple prefijo `/` en el cuadro de entrada del chat. Estos comandos ayudan a estandarizar procesos en tu equipo y hacen más eficientes las tareas comunes.

<Info>
  Los comandos están actualmente en versión beta. La función y la sintaxis pueden cambiar a medida que seguimos mejorándola.
</Info>

<div id="how-commands-work">
  ## Cómo funcionan los comandos
</div>

Los comandos se definen como archivos Markdown de texto plano que se pueden guardar en dos ubicaciones:

1. **Comandos del proyecto**: almacenados en el directorio `.cursor/commands` de tu proyecto
2. **Comandos globales**: almacenados en el directorio `~/.cursor/commands` de tu directorio de inicio

Cuando escribes `/` en el cuadro de entrada del chat, Cursor detecta y muestra automáticamente los comandos disponibles de ambos directorios, haciéndolos accesibles al instante en todo tu flujo de trabajo.

<div id="creating-commands">
  ## Creación de comandos
</div>

1. Crea un directorio `.cursor/commands` en la raíz de tu proyecto
2. Añade archivos `.md` con nombres descriptivos (p. ej., `review-code.md`, `write-tests.md`)
3. Escribe contenido en Markdown sencillo que describa lo que debe hacer el comando
4. Los comandos aparecerán automáticamente en el chat cuando escribas `/`

Aquí tienes un ejemplo de cómo podría verse la estructura de tu directorio de comandos:

```
.cursor/
└── commands/
    ├── atender-comentarios-de-pr-de-github.md
    ├── lista-de-verificación-para-revisión-de-código.md
    ├── crear-pr.md
    ├── revisión-ligera-de-diffs-existentes.md
    ├── incorporación-de-nuevo-desarrollador.md
    ├── ejecutar-todas-las-pruebas-y-corregir.md
    ├── auditoría-de-seguridad.md
    └── configurar-nueva-función.md
```

<div id="examples">
  ## Ejemplos
</div>

Prueba estos comandos en tus proyectos para hacerte una idea de cómo funcionan.

<AccordionGroup>
  <Accordion title="Lista de verificación para revisión de código">
    ```markdown  theme={null}
    # Lista de verificación para revisión de código

## Descripción general
    Lista de verificación completa para realizar revisiones de código exhaustivas y garantizar calidad, seguridad y mantenibilidad.

## Categorías de revisión

### Funcionalidad
    - [ ] El código hace lo que debe
    - [ ] Se contemplan los casos límite
    - [ ] El manejo de errores es adecuado
    - [ ] No hay errores obvios ni fallos de lógica

### Calidad del código
    - [ ] El código es legible y está bien estructurado
    - [ ] Las funciones son pequeñas y enfocadas
    - [ ] Los nombres de variables son descriptivos
    - [ ] No hay duplicación de código
    - [ ] Sigue las convenciones del proyecto

### Seguridad
    - [ ] No hay vulnerabilidades de seguridad evidentes
    - [ ] Hay validación de entradas
    - [ ] Los datos sensibles se manejan correctamente
    - [ ] No hay secretos codificados en duro
    ```
  </Accordion>

<Accordion title="Auditoría de seguridad">
    ```markdown  theme={null}
    # Auditoría de seguridad

## Descripción general
    Revisión de seguridad integral para identificar y corregir vulnerabilidades en la base de código.

## Pasos
    1. **Auditoría de dependencias**
       - Detectar vulnerabilidades conocidas
       - Actualizar paquetes desactualizados
       - Revisar dependencias de terceros

2. **Revisión de seguridad del código**
       - Detectar vulnerabilidades comunes
       - Revisar autenticación/autorización
       - Auditar prácticas de manejo de datos

3. **Seguridad de la infraestructura**
       - Revisar variables de entorno
       - Verificar controles de acceso
       - Auditar la seguridad de la red

## Lista de verificación de seguridad
    - [ ] Dependencias actualizadas y seguras
    - [ ] Sin secretos hardcodeados
    - [ ] Validación de entradas implementada
    - [ ] Autenticación segura
    - [ ] Autorización configurada correctamente
    ```
  </Accordion>

<Accordion title="Configurar nueva función">
    ```markdown  theme={null}
    # Configurar nueva feature

## Descripción general
    Configura sistemáticamente una nueva feature, desde la planificación inicial hasta la estructura de implementación.

## Pasos
    1. **Definir requisitos**
       - Aclarar el alcance y los objetivos de la feature
       - Identificar historias de usuario y criterios de aceptación
       - Planificar el enfoque técnico

2. **Crear rama de la feature**
       - Ramificar desde main/develop
       - Configurar el entorno de desarrollo local
       - Configurar cualquier dependencia nueva

3. **Planificar la arquitectura**
       - Diseñar modelos de datos y APIs
       - Planificar componentes de UI y flujos
       - Definir la estrategia de testing

## Checklist de configuración de la feature
    - [ ] Requisitos documentados
    - [ ] Historias de usuario redactadas
    - [ ] Enfoque técnico definido
    - [ ] Rama de la feature creada
    - [ ] Entorno de desarrollo listo
    ```
  </Accordion>

<Accordion title="Crear pull request">
    ```markdown  theme={null}
    # Crear PR

## Descripción general
    Crea un pull request bien estructurado con una descripción clara, etiquetas y revisores.

## Pasos
    1. **Preparar la rama**
       - Asegúrate de que todos los cambios estén confirmados
       - Envía la rama al remoto
       - Verifica que la rama esté actualizada con main

2. **Escribir la descripción del PR**
       - Resume los cambios de forma clara
       - Incluye contexto y motivación
       - Enumera cualquier cambio incompatible
       - Agrega capturas de pantalla si hay cambios en la UI

3. **Configurar el PR**
       - Crea el PR con un título descriptivo
       - Agrega etiquetas adecuadas
       - Asigna revisores
       - Vincula issues relacionados

## Plantilla de PR
    - [ ] Funcionalidad A
    - [ ] Corrección de bug B
    - [ ] Pruebas unitarias aprobadas
    - [ ] Pruebas manuales completadas
    ```
  </Accordion>

<Accordion title="Ejecuta las pruebas y corrige los fallos">
    ```markdown  theme={null}
    # Ejecuta todas las pruebas y corrige los errores

## Descripción general
    Ejecuta toda la batería de pruebas y corrige de forma sistemática cualquier error, garantizando la calidad y la funcionalidad del código.

## Pasos
    1. **Ejecuta la batería de pruebas**
       - Ejecuta todas las pruebas del proyecto
       - Captura la salida e identifica los errores
       - Revisa tanto las pruebas unitarias como las de integración

2. **Analiza los errores**
       - Clasifícalos por tipo: intermitentes, rotos, nuevos
       - Prioriza las correcciones según el impacto
       - Revisa si los errores están relacionados con cambios recientes

3. **Corrige los problemas de forma sistemática**
       - Empieza por los errores más críticos
       - Corrige un problema a la vez
       - Vuelve a ejecutar las pruebas después de cada corrección
    ```
  </Accordion>

<Accordion title="Incorporar a un nuevo developer">
    ```markdown  theme={null}
    # Incorporar a un nuevo developer

## Descripción general
    Proceso completo de onboarding para que un developer nuevo arranque y quede listo rápido.

## Pasos
    1. **Configuración del entorno**
       - Instalar las herramientas requeridas
       - Configurar el entorno de desarrollo
       - Configurar el IDE y las extensiones
       - Configurar git y las claves SSH

2. **Familiarización con el proyecto**
       - Revisar la estructura del proyecto
       - Entender la arquitectura
       - Leer la documentación clave
       - Configurar la base de datos local

## Lista de verificación de onboarding
    - [ ] Entorno de desarrollo listo
    - [ ] Todas las pruebas pasando
    - [ ] Puedes ejecutar la app localmente
    - [ ] Base de datos configurada y funcionando
    - [ ] Primer PR enviado
    ```
  </Accordion>
</AccordionGroup>

# Compacto
Source: https://docs.cursor.com/es/agent/chat/compact

Ahorra espacio en el chat con la interfaz en modo compacto

El modo compacto ofrece una interfaz de chat más limpia al reducir el ruido visual y aprovechar al máximo el espacio para las conversaciones.

<div id="overview">
  ## Descripción general
</div>

Cuando está activado, el modo compacto transforma la interfaz del chat al:

* **Ocultar iconos** para una apariencia más limpia y minimalista
* **Plegar automáticamente los diffs** para reducir el ruido visual
* **Plegar automáticamente el campo de entrada** para maximizar el espacio de la conversación

Esta configuración es especialmente útil cuando trabajas en pantallas más pequeñas o cuando prefieres una experiencia de chat enfocada y sin distracciones.

<div id="before-and-after">
  ## Antes y después
</div>

<div id="default-mode">
  ### Modo predeterminado
</div>

<div id="compact-mode">
  ### Modo compacto
</div>

<div id="enabling-compact-mode">
  ## Activar el modo compacto
</div>

Para activar el modo compacto:

1. Abre la configuración de Cursor
2. Ve a la sección de **Chat**
3. Activa **Compact Mode** para habilitarlo

La interfaz se actualizará de inmediato a una vista más compacta, dándote más espacio para concentrarte en tus conversaciones.

# Duplicar
Source: https://docs.cursor.com/es/agent/chat/duplicate

Crea ramas desde cualquier punto de una conversación

Duplica o crea forks de chats para explorar soluciones alternativas sin perder tu conversación actual.

<div id="how-to-duplicate">
  ## Cómo duplicar
</div>

1. Encuentra dónde quieres crear una rama
2. Haz clic en los tres puntos del mensaje
3. Selecciona "Duplicate Chat"

* Se mantiene el contexto hasta ese punto
* La conversación original no cambia
* Ambos chats conservan historiales separados

# Exportar
Source: https://docs.cursor.com/es/agent/chat/export

Exportar chats en formato Markdown

Exporta los chats del Agent como archivos Markdown para compartir o documentar.

<div id="whats-exported">
  ## Qué se exporta
</div>

* Todos los mensajes y respuestas
* Bloques de código con resaltado de sintaxis
* Referencias a archivos y contexto
* Flujo cronológico de la conversación

<div id="how-to-export">
  ## Cómo exportar
</div>

1. Ve al chat que quieres exportar
2. Haz clic en el menú contextual → "Export Chat"
3. Guarda el archivo localmente

<Warning>
  Revisa las exportaciones en busca de datos sensibles: claves de API, URL internas, código propietario,
  información personal
</Warning>

# Historial
Source: https://docs.cursor.com/es/agent/chat/history

Ver y gestionar conversaciones de chat

Accede a conversaciones anteriores del agente desde el panel de historial.

<div id="opening-history">
  ## Abrir el historial
</div>

* Haz clic en el icono de historial en el panel lateral de Agent
* Presiona <Kbd tooltip="Open chat history">Opt Cmd '</Kbd>

<div id="managing-chats">
  ## Administrar chats
</div>

* **Editar títulos**: Haz clic para cambiar el nombre
* **Eliminar**: Borra los chats que no necesitas
* **Abrir**: Haz clic para revisar la conversación completa

El historial de chats se guarda localmente en una base de datos SQLite en tu equipo.

<Note>
  Para conservar los chats, [expórtalos](/es/agent/chats/export) como Markdown.
</Note>

<div id="background-agents">
  ## Agentes en segundo plano
</div>

Los chats de agentes en segundo plano no aparecen en el historial habitual; en su lugar, se guardan en una base de datos remota. Usa <Kbd tooltip="Open background agent control panel">Cmd E</Kbd> para verlos.

<div id="referencing-past-chats">
  ## Referenciar chats anteriores
</div>

Usa [@Past Chats](/es/context/@-symbols/@-past-chats) para incluir contexto de conversaciones anteriores en tu chat actual.

# Resumen
Source: https://docs.cursor.com/es/agent/chat/summarization

Gestión del contexto para conversaciones largas en el chat

<div id="message-summarization">
  ## Resumen de mensajes
</div>

A medida que las conversaciones crecen, Cursor resume y gestiona el contexto automáticamente para mantener tus chats eficientes. Aprende a usar el menú contextual y a entender cómo se condensan los archivos para ajustarse a las ventanas de contexto del modelo.

<div id="using-the-summarize-command">
  ### Usar el comando /summarize
</div>

Podés activar manualmente un resumen usando el comando `/summarize` en el chat. Este comando ayuda a gestionar el contexto cuando las conversaciones se hacen demasiado largas, así podés seguir trabajando de forma eficiente sin perder información importante.

<Info>
  Para profundizar en cómo funciona el contexto en Cursor, mirá nuestra guía [Working with
  Context](/es/guides/working-with-context).
</Info>

<div id="how-summarization-works">
  ### Cómo funciona el resumen
</div>

Cuando las conversaciones se alargan, superan el límite de la ventana de contexto del modelo:

<div className="bg-white px-3 py-2 mb-2 rounded border border-dashed border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>
    </div>
  </div>
</Frame>

Para resolver esto, Cursor resume los mensajes más antiguos para hacer espacio a conversaciones nuevas.

<div className="bg-neutral-100 px-3 py-2 mb-2 rounded border border-dashed border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">
          Mensajes resumidos
        </div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>

<div className="bg-white px-3 py-2 mb-2 rounded border border-neutral-400 w-full">
        <div className="text-xs text-neutral-700 font-medium">Cursor</div>
      </div>
    </div>
  </div>
</Frame>

<div id="file-folder-condensation">
  ## Condensación de archivos y carpetas
</div>

Aunque el resumen del chat se encarga de conversaciones largas, Cursor usa una estrategia distinta para manejar archivos y carpetas grandes: **condensación inteligente**. Cuando incluyes archivos en tu conversación, Cursor decide la mejor forma de presentarlos según su tamaño y el espacio de contexto disponible.

Estos son los diferentes estados en los que puede estar un archivo o carpeta:

<div id="condensed">
  ### Resumen
</div>

Cuando los archivos o carpetas son demasiado grandes para entrar en la ventana de contexto, Cursor los condensa automáticamente. Al condensar, el modelo ve elementos estructurales clave como firmas de funciones, clases y métodos. Desde esta vista condensada, el modelo puede elegir expandir archivos específicos si hace falta. Este enfoque maximiza el uso efectivo de la ventana de contexto disponible.

<div id="significantly-condensed">
  ### Significativamente comprimido
</div>

Cuando un nombre de archivo aparece con la etiqueta “Significantly Condensed”, significa que el archivo es demasiado grande para incluirlo completo, incluso en una versión comprimida. Solo se mostrará el nombre del archivo al modelo.

<div id="not-included">
  ### No incluido
</div>

Cuando aparece un icono de advertencia junto a un archivo o carpeta, el elemento es demasiado grande para incluirlo en la ventana de contexto, incluso en versión condensada. Esto te ayuda a entender qué partes de tu codebase son accesibles para el modelo.

# Pestañas
Source: https://docs.cursor.com/es/agent/chat/tabs

Ejecuta varias conversaciones del Agente a la vez

<div id="overview">
  ## Descripción general
</div>

Presiona <Kbd>Cmd+T</Kbd> para crear pestañas nuevas. Cada pestaña mantiene por separado el historial de la conversación, el contexto y la selección de modelo.

<Tip>
  Para trabajar en paralelo, prueba los [Background Agents](/es/background-agents)
</Tip>

<div id="managing-tabs">
  ## Gestión de pestañas
</div>

* Creá nuevas pestañas con <Kbd>Cmd+T</Kbd>. Cada pestaña inicia una conversación nueva y mantiene su propio contexto.

* Cambiá entre pestañas haciendo clic en sus encabezados o usando <Kbd>Ctrl+Tab</Kbd> para recorrerlas.

* Los títulos de las pestañas se generan automáticamente después del primer mensaje, pero podés renombrarlos con el botón derecho en el encabezado de la pestaña.

<Tip>
  Usá una tarea por pestaña, brindá descripciones iniciales claras y cerrá las
  pestañas terminadas para mantener tu espacio de trabajo organizado.
</Tip>

<div id="conflicts">
  ### Conflictos
</div>

Cursor evita que varias pestañas editen los mismos archivos. Se te pedirá que resuelvas los conflictos.

<div id="reference-other-chats">
  ## Referenciar otros chats
</div>

Usa [@Past Chats](/es/context/@-symbols/@-past-chats) para incluir contexto de otras pestañas o sesiones anteriores.

# Modos
Source: https://docs.cursor.com/es/agent/modes

Elige el modo adecuado para tu tarea: de codificación autónoma a ediciones enfocadas

Agent ofrece distintos modos optimizados para tareas específicas. Cada modo tiene diferentes capacidades y herramientas activadas para adaptarse a tu flujo de trabajo.

<div id="agent">
  ## Agent
</div>

El modo predeterminado para tareas de programación complejas. Agent explora tu base de código de forma autónoma, edita varios archivos, ejecuta comandos y corrige errores para completar tus solicitudes.

Modo de solo lectura para aprender y explorar. Ask busca en tu código y te da respuestas sin hacer ningún cambio: perfecto para entender el código antes de modificarlo.

<div id="custom">
  ## Personalizado
</div>

Crea tus propios modos con combinaciones específicas de herramientas e instrucciones. Mezcla capacidades según lo que mejor encaje con tu flujo de trabajo.

<Note>
  Los modos personalizados están en beta. Actívalos en `Cursor Settings` → `Chat` → `Custom
      Modes`
</Note>

<div id="examples">
  ### Ejemplos
</div>

<AccordionGroup>
  <Accordion title="Learn">
    **Herramientas:** All Search\
    **Instrucciones:** Enfócate en explicar los conceptos a fondo y haz preguntas para aclarar
  </Accordion>

{" "}

<Accordion title="Refactor">
    **Herramientas:** Edit & Reapply **Instrucciones:** Mejora la estructura del código sin
    agregar nueva funcionalidad
  </Accordion>

{" "}

<Accordion title="Plan">
    **Herramientas:** Codebase, Read file, Terminal **Instrucciones:** Crea planes de implementación detallados en `plan.md`
  </Accordion>

<Accordion title="Debug">
    **Herramientas:** All Search, Terminal, Edit & Reapply\
    **Instrucciones:** Investiga a fondo los problemas antes de proponer soluciones
  </Accordion>
</AccordionGroup>

<div id="switching-modes">
  ## Cambiar de modo
</div>

* Usa el menú desplegable del selector de modo en Agent
* Presiona <Kbd>Cmd+.</Kbd> para cambiar rápidamente
* Configura los atajos de teclado en [configuración](#settings)

<div id="settings">
  ## Configuración
</div>

Todos los modos comparten opciones de configuración comunes:

Configuraciones específicas por modo:

# Descripción general
Source: https://docs.cursor.com/es/agent/overview

Asistente para tareas de programación autónomas, comandos de terminal y edición de código

Agent es el asistente de Cursor que puede completar tareas de programación complejas por sí solo, ejecutar comandos en la terminal y editar código. Accede desde el panel lateral con <Kbd>Cmd+I</Kbd>.

<div className="mt-24 flex flex-col gap-12">
  <Columns className="gap-4">
    <div>
      <h2 className="text-lg font-medium mb-2">
        <a href="/es/agent/modes" className="hover:text-primary transition-colors">
          Modos
        </a>
      </h2>

<p className="text-sm">
        Elegí entre Agent, Ask o creá modos personalizados. Cada modo ofrece
        distintas capacidades y herramientas para adaptarse a tu flujo de trabajo.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/tools" className="hover:text-primary transition-colors">
          Herramientas
        </a>
      </h3>

<p className="text-sm">
        Agent utiliza herramientas para buscar, editar y ejecutar comandos. Desde la
        búsqueda semántica en la base de código hasta la ejecución en la terminal, estas
        herramientas permiten completar tareas de forma autónoma.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/apply" className="hover:text-primary transition-colors">
          Aplicar cambios
        </a>
      </h3>

<p className="text-sm">
        Integra bloques de código sugeridos por IA en tu codebase. Apply gestiona
        cambios a gran escala de forma eficiente sin perder precisión.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/review" className="hover:text-primary transition-colors">
          Revisar diffs
        </a>
      </h3>

<p className="text-sm">
        Revisa los cambios antes de aceptarlos. La interfaz de revisión muestra añadidos
        y eliminaciones con líneas codificadas por color para que tengas control sobre las modificaciones.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/chats/tabs" className="hover:text-primary transition-colors">
          Pestañas de chat
        </a>
      </h3>

<p className="text-sm">
        Lleva varias conversaciones a la vez con <Kbd>Cmd+T</Kbd>. Cada pestaña
        mantiene su propio contexto, historial y modelo seleccionado.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/chats/checkpoints" className="hover:text-primary transition-colors">
          Checkpoints
        </a>
      </h3>

<p className="text-sm">
        Las instantáneas automáticas registran los cambios del Agent. Restaura estados anteriores si
        los cambios no funcionan como esperabas o para probar enfoques distintos.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/terminal" className="hover:text-primary transition-colors">
          Integración con la terminal
        </a>
      </h3>

<p className="text-sm">
        Agent ejecuta comandos en la terminal, supervisa la salida y gestiona procesos
        de varios pasos. Configura la ejecución automática para flujos de trabajo de
        confianza o pide confirmación por seguridad.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/chats/history" className="hover:text-primary transition-colors">
          Historial de chat
        </a>
      </h3>

<p className="text-sm">
        Accede a conversaciones anteriores con <Kbd>Opt Cmd '</Kbd>. Revisa
        chats previos, sigue tus sesiones de código y consulta el contexto de
        conversaciones anteriores.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/agent/chats/export" className="hover:text-primary transition-colors">
          Exportar chats
        </a>
      </h3>

<p className="text-sm">
        Exporta conversaciones en formato Markdown. Comparte soluciones con tu equipo, documenta decisiones o crea bases de conocimiento a partir de sesiones de programación.
      </p>
    </div>

<Columns className="gap-4">
    <div>
      <h3 className="text-lg font-medium mb-2">
        <a href="/es/context/rules" className="hover:text-primary transition-colors">
          Reglas
        </a>
      </h3>

<p className="text-sm">
        Define instrucciones personalizadas para el comportamiento de Agent. Las reglas ayudan a mantener estándares de codificación, aplicar patrones y personalizar cómo Agent te asiste en tu proyecto.
      </p>
    </div>

# Planificación
Source: https://docs.cursor.com/es/agent/planning

Cómo Agent planifica y gestiona tareas complejas con to-dos y colas

Agent puede planificar con anticipación y gestionar tareas complejas con listas de tareas estructuradas y cola de mensajes, lo que facilita entender y hacer seguimiento de tareas de largo plazo.

<div id="agent-to-dos">
  ## Tareas pendientes de Agent
</div>

Agent puede dividir tareas largas en pasos manejables con dependencias, creando un plan estructurado que se actualiza conforme avanza el trabajo.

<div id="how-it-works">
  ### Cómo funciona
</div>

* Agent crea automáticamente listas de pendientes para tareas complejas
* Cada ítem puede depender de otras tareas
* La lista se actualiza en tiempo real a medida que avanza el trabajo
* Las tareas completadas se marcan automáticamente

<div id="visibility">
  ### Visibilidad
</div>

* Las tareas pendientes aparecen en la interfaz del chat
* Si la [integración con Slack](/es/slack) está configurada, las tareas también se ven ahí
* Podés ver el desglose completo de la tarea en cualquier momento

<Tip>
  Para planificar mejor, describí claramente tu objetivo final. Agent creará desgloses de tareas más
  precisos cuando entienda el alcance completo.
</Tip>

<Note>La planificación y las tareas pendientes no están disponibles actualmente en el modo auto.</Note>

<div id="queued-messages">
  ## Mensajes en cola
</div>

Pon mensajes de seguimiento en cola mientras Agent trabaja en la tarea actual. Tus instrucciones esperan su turno y se ejecutan automáticamente cuando estén listas.

1. Mientras Agent está trabajando, escribe tu siguiente instrucción
2. Pulsa <Kbd>Ctrl+Enter</Kbd> para añadirla a la cola
3. Los mensajes aparecen en orden debajo de la tarea activa
4. Reordena los mensajes en cola haciendo clic en la flecha
5. Agent los procesa secuencialmente al terminar

<div id="override-the-queue">
  ### Omitir la cola
</div>

Para poner tu mensaje en cola en lugar de usar la mensajería predeterminada, usa <Kbd>Ctrl+Enter</Kbd>. Para enviar un mensaje de inmediato sin ponerlo en cola, usa <Kbd>Cmd+Enter</Kbd>. Esto “empuja” tu mensaje a la fuerza, omitiendo la cola para ejecutarlo de inmediato.

<div id="default-messaging">
  ## Mensajería predeterminada
</div>

De forma predeterminada, los mensajes se envían lo más rápido posible, generalmente apareciendo justo después de que Agent complete una llamada a una herramienta. Esto ofrece la experiencia más ágil.

<div id="how-default-messaging-works">
  ### Cómo funciona la mensajería predeterminada
</div>

* Tu mensaje se añade al mensaje de usuario más reciente en el chat
* Los mensajes suelen adjuntarse a los resultados de las herramientas y se envían en cuanto están listos
* Esto crea un flujo de conversación más natural sin interrumpir el trabajo actual de Agent
* Por defecto, esto ocurre cuando presionas Enter mientras Agent está trabajando

# Diffs y revisión
Source: https://docs.cursor.com/es/agent/review

Revisa y gestiona los cambios de código generados por el agente de IA

Cuando Agent genera cambios de código, se muestran en una interfaz de revisión que marca adiciones y eliminaciones con líneas codificadas por color. Esto te permite revisar y controlar qué cambios se aplican a tu base de código.

La interfaz de revisión muestra los cambios de código en un formato diff familiar:

<div id="diffs">
  ## Diffs
</div>

<div className="full-width-table">
  | Tipo                   | Significado                      | Ejemplo                                                                                               |
  | :--------------------- | :------------------------------- | :---------------------------------------------------------------------------------------------------- |
  | **Líneas añadidas**    | Nuevas incorporaciones de código | <code className="bg-green-100 text-green-800 px-2 py-1 rounded">+ const newVariable = 'hello';</code> |
  | **Líneas eliminadas**  | Eliminaciones de código          | <code className="bg-red-100 text-red-800 px-2 py-1 rounded">- const oldVariable = 'goodbye';</code>   |
  | **Líneas de contexto** | Código circundante sin cambios   | <code className="bg-gray-100 text-gray-600 px-2 py-1 rounded"> function example() {}</code>           |
</div>

<div id="review">
  ## Revisión
</div>

Cuando termine la generación, vas a ver un aviso para revisar todos los cambios antes de continuar. Esto te da una vista general de lo que se va a modificar.

<div id="file-by-file">
  ### Archivo por archivo
</div>

Aparece una barra de revisión flotante en la parte inferior de tu pantalla que te permite:

* **Aceptar** o **rechazar** cambios del archivo actual
* Ir al **siguiente archivo** con cambios pendientes
  <Frame>
    <video src="https://mintcdn.com/cursor/BfJOqJ1Wb8EvuXyr/images/chat/review/review-bar.mp4?fit=max&auto=format&n=BfJOqJ1Wb8EvuXyr&q=85&s=5dca0fe7aba3c79e6760cb264821a617" autoPlay loop muted playsInline controls data-path="images/chat/review/review-bar.mp4">
      Tu navegador no admite la etiqueta de video.
    </video>
  </Frame>

<div id="selective-acceptance">
  ### Aceptación selectiva
</div>

Para un control más preciso:

* Para aceptar la mayoría de los cambios: rechazá las líneas que no querés y después hacé clic en **Aceptar todo**
* Para rechazar la mayoría de los cambios: aceptá las líneas que sí querés y después hacé clic en **Rechazar todo**

<div id="review-changes">
  ## Revisar cambios
</div>

Al final de la respuesta del agente, haz clic en el botón **Revisar cambios** para ver el diff completo de las modificaciones.

# Terminal
Source: https://docs.cursor.com/es/agent/terminal

Ejecuta comandos de la terminal automáticamente como parte de las operaciones del agente

El agente ejecuta comandos en la terminal nativa de Cursor y conserva el historial. Haz clic en Skip para enviar <kbd>Ctrl+C</kbd> y detener los comandos.

<div id="troubleshooting">
  ## Solución de problemas
</div>

<Info>
  Algunos temas de shell (por ejemplo, Powerlevel9k/Powerlevel10k) pueden interferir con
  la salida del terminal en línea. Si la salida de tu comando aparece truncada o
  con un formato incorrecto, desactiva el tema o cambia a un prompt más simple cuando se ejecute Agent.
</Info>

<div id="disable-heavy-prompts-for-agent-sessions">
  ### Desactiva prompts pesados para sesiones de Agent
</div>

Usa la variable de entorno `CURSOR_AGENT` en la configuración de tu shell para detectar cuándo
se está ejecutando Agent y saltarte la inicialización de prompts/temas vistosos.

```zsh  theme={null}
# ~/.zshrc — desactiva Powerlevel10k cuando se ejecute Cursor Agent
if [[ -n "$CURSOR_AGENT" ]]; then
  # Omitir la inicialización del tema para mejorar la compatibilidad
else
  [[ -r ~/.p10k.zsh ]] && source ~/.p10k.zsh
fi
```

```bash  theme={null}
# ~/.bashrc — usar un prompt simple en sesiones del Agent
if [[ -n "$CURSOR_AGENT" ]]; then
  PS1='\u@\h \W \$ '
fi
```

# Herramientas
Source: https://docs.cursor.com/es/agent/tools

Herramientas disponibles para que los agentes busquen, editen y ejecuten código

Una lista de todas las herramientas disponibles para los modos dentro de [Agent](/es/agent/overview), que podés habilitar o deshabilitar al crear tus propios [modos personalizados](/es/agent/modes#custom).

<Note>
  No hay límite en la cantidad de llamadas a herramientas que Agent puede hacer durante una tarea. Agent seguirá usando herramientas según sea necesario para completar tu pedido.
</Note>

<div id="search">
  ## Búsqueda
</div>

Herramientas para buscar en tu código y en la web y encontrar información relevante.

<AccordionGroup>
  <Accordion title="Read File" icon="file-lines">
    Lee hasta 250 líneas (750 en modo máximo) de un archivo.
  </Accordion>

<Accordion title="List Directory" icon="folder-open">
    Lee la estructura de un directorio sin leer el contenido de los archivos.
  </Accordion>

<Accordion title="Codebase" icon="database">
    Realiza búsquedas semánticas en tu [código
    indexado](/es/context/codebase-indexing).
  </Accordion>

<Accordion title="Grep" icon="magnifying-glass">
    Busca palabras clave o patrones exactos dentro de archivos.
  </Accordion>

<Accordion title="Search Files" icon="file-magnifying-glass">
    Encuentra archivos por nombre usando coincidencia difusa.
  </Accordion>

<Accordion title="Web" icon="globe">
    Genera consultas y realiza búsquedas en la web.
  </Accordion>

<Accordion title="Fetch Rules" icon="gavel">
    Obtén [reglas](/es/context/rules) específicas según el tipo y la descripción.
  </Accordion>
</AccordionGroup>

<div id="edit">
  ## Editar
</div>

Herramientas para hacer cambios específicos en tus archivos y tu código.

<AccordionGroup>
  <Accordion title="Editar y reaplicar" icon="pencil">
    Sugiere cambios en archivos y [aplícalos](/es/agent/apply) automáticamente.
  </Accordion>

<Accordion title="Eliminar archivo" icon="trash">
    Elimina archivos de forma autónoma (puedes desactivarlo en la configuración).
  </Accordion>
</AccordionGroup>

<div id="run">
  ## Ejecutar
</div>

Chat puede interactuar con tu terminal.

<AccordionGroup>
  <Accordion title="Terminal" icon="terminal">
    Ejecuta comandos en la terminal y monitorea la salida.
  </Accordion>
</AccordionGroup>

<Note>De forma predeterminada, Cursor usa el primer perfil de terminal disponible.</Note>

Para establecer tu perfil de terminal preferido:

1. Abre la paleta de comandos (`Cmd/Ctrl+Shift+P`)
2. Busca "Terminal: Select Default Profile"
3. Elige el perfil que quieras

Chat puede usar servidores MCP configurados para interactuar con servicios externos, como bases de datos o API de terceros.

<AccordionGroup>
  <Accordion title="Activar/desactivar servidores MCP" icon="server">
    Activa o desactiva los servidores MCP disponibles. Respeta la configuración de ejecución automática.
  </Accordion>
</AccordionGroup>

Aprende más sobre el [Model Context Protocol](/es/context/model-context-protocol) y explora los servidores disponibles en el [directorio de MCP](/es/tools).

<div id="advanced-options">
  ## Opciones avanzadas
</div>

<AccordionGroup>
  <Accordion title="Aplicar ediciones automáticamente" icon="check">
    Aplica las ediciones automáticamente sin confirmación manual.
  </Accordion>

<Accordion title="Ejecución automática" icon="play">
    Ejecuta automáticamente comandos de terminal y acepta ediciones. Útil para ejecutar suites de pruebas y verificar cambios.

<Accordion title="Guardrails" icon="shield">
    Configura listas de permitidos para especificar qué herramientas pueden ejecutarse automáticamente. Estas listas mejoran la seguridad al definir explícitamente las operaciones permitidas.
  </Accordion>

<Accordion title="Corregir errores automáticamente" icon="wrench">
    Resuelve automáticamente los errores y advertencias del linter cuando el Agent los encuentre.
  </Accordion>
</AccordionGroup>

# Agentes en segundo plano
Source: https://docs.cursor.com/es/background-agent

Agentes remotos asíncronos en Cursor

Con los agentes en segundo plano, podés lanzar agentes asíncronos que editen y ejecuten código en un entorno remoto. Mirá su estado, mandá seguimientos o tomá el control en cualquier momento.

Podés acceder a los agentes en segundo plano de dos maneras:

1. **Barra lateral de agentes en segundo plano**: Usá la pestaña de agentes en segundo plano en la barra lateral nativa de Cursor para ver todos los agentes en segundo plano asociados con tu cuenta, buscar agentes existentes y arrancar nuevos.
2. **Modo de agente en segundo plano**: Presioná <Kbd tooltip="Trigger background agent mode">Cmd E</Kbd> para activar el modo de agente en segundo plano en la UI.

Después de enviar un prompt, seleccioná tu agente de la lista para ver el estado y entrar a la máquina.

<Note>
  <p className="!mb-0">
    Los agentes en segundo plano requieren retener datos por unos pocos días.
  </p>
</Note>

<div id="setup">
  ## Configuración
</div>

Por defecto, los agentes en segundo plano se ejecutan en una máquina aislada basada en Ubuntu. Tienen acceso a Internet y pueden instalar paquetes.

<div id="github-connection">
  #### Conexión con GitHub
</div>

Los agentes en segundo plano clonan tu repositorio desde GitHub y trabajan en una rama aparte, haciendo push a tu repositorio para facilitar la entrega.

Concede permisos de lectura y escritura a tu repositorio (y a cualquier repositorio dependiente o submódulo). En el futuro, admitiremos otros proveedores (GitLab, Bitbucket, etc.).

<div id="ip-allow-list-configuration">
  ##### Configuración de la lista de IP permitidas
</div>

Si tu organización usa la función de lista de IP permitidas de GitHub, vas a tener que configurar el acceso para los agentes en segundo plano. Consulta la [documentación de la integración con GitHub](/es/integrations/github#ip-allow-list-configuration) para ver las instrucciones completas de configuración, incluida la información de contacto y las direcciones IP.

<div id="base-environment-setup">
  #### Configuración base del entorno
</div>

Para casos avanzados, configura el entorno por tu cuenta. Consigue una instancia del IDE conectada a la máquina remota. Prepara tu máquina, instala herramientas y paquetes, y luego toma un snapshot. Configura los ajustes de runtime:

* El comando de instalación se ejecuta antes de que arranque un agente e instala las dependencias de runtime. Esto puede implicar ejecutar `npm install` o `bazel build`.
* Los terminales ejecutan procesos en segundo plano mientras el agente trabaja, como iniciar un servidor web o compilar archivos protobuf.

Para los casos más avanzados, usa un Dockerfile para configurar la máquina. El Dockerfile te permite configurar dependencias a nivel del sistema: instalar versiones específicas de compiladores, depuradores o cambiar la imagen base del SO. No hagas `COPY` de todo el proyecto: nosotros gestionamos el workspace y hacemos checkout del commit correcto. Aun así, gestiona la instalación de dependencias en el script de instalación.

Ingresa cualquier secreto requerido para tu entorno de desarrollo: se almacenan cifrados en reposo (usando KMS) en nuestra base de datos y se proporcionan en el entorno del agente en segundo plano.

La configuración de la máquina vive en `.cursor/environment.json`, que puedes commitear en tu repo (recomendado) o almacenar de forma privada. El flujo de configuración te guía para crear `environment.json`.

<div id="maintenance-commands">
  #### Comandos de mantenimiento
</div>

Al configurar una máquina nueva, empezamos desde el entorno base y luego ejecutamos el comando `install` de tu `environment.json`. Este comando es el que ejecutaría un desarrollador al cambiar de rama: instalar cualquier dependencia nueva.

Para la mayoría, el comando `install` es `npm install` o `bazel build`.

Para asegurar un arranque rápido de la máquina, almacenamos en caché el estado del disco después de que se ejecute el comando `install`. Diseñalo para que pueda ejecutarse varias veces. Solo persiste el estado del disco desde el comando `install`: los procesos iniciados aquí no seguirán en ejecución cuando el agente arranque.

<div id="startup-commands">
  #### Comandos de inicio
</div>

Después de ejecutar `install`, la máquina arranca y ejecutamos el comando `start`, seguido de iniciar cualquier `terminals`. Esto levanta procesos que deberían estar en ejecución cuando se ejecute el agente.

El comando `start` a menudo se puede omitir. Úsalo si tu entorno de dev depende de Docker: pon `sudo service docker start` en el comando `start`.

Los `terminals` son para código de la app. Estos terminales se ejecutan en una sesión de `tmux` disponible para ti y para el agente. Por ejemplo, muchos repos de sitios web ponen `npm run watch` como un terminal.

<div id="the-environmentjson-spec">
  #### La especificación de `environment.json`
</div>

El archivo `environment.json` puede tener este aspecto:

```json  theme={null}
{
  "snapshot": "POPULATED_FROM_SETTINGS",
  "install": "npm install",
  "terminals": [
    {
      "name": "Iniciar Next.js",
      "command": "npm run dev"
    }
  ]
}
```

Formalmente, la especificación se [define aquí](https://www.cursor.com/schemas/environment.schema.json).

<div id="models">
  ## Modelos
</div>

Solo los modelos compatibles con [Max Mode](/es/context/max-mode) están disponibles para los agentes en segundo plano.

<div id="pricing">
  ## Precios
</div>

Conocé más sobre los [precios de Background Agent](/es/account/pricing#background-agent).

<div id="security">
  ## Seguridad
</div>

Los Background Agents están disponibles en Privacy Mode. Nunca entrenamos con tu código y solo lo conservamos para ejecutar el agente. [Más información sobre Privacy Mode](https://www.cursor.com/privacy-overview).

Lo que deberías saber:

1. Concede privilegios de lectura y escritura a nuestra app de GitHub en los repos que quieras editar. Usamos esto para clonar el repo y hacer cambios.
2. Tu código se ejecuta dentro de nuestra infraestructura en AWS, en VMs aisladas, y se almacena en discos de VM mientras el agente está activo.
3. El agente tiene acceso a Internet.
4. El agente ejecuta automáticamente todos los comandos de terminal, lo que le permite iterar en las pruebas. Esto difiere del foreground agent, que requiere tu aprobación para cada comando. La ejecución automática introduce riesgo de exfiltración de datos: atacantes podrían lanzar ataques de prompt injection, engañando al agente para que suba código a sitios web maliciosos. Consulta la [explicación de OpenAI sobre los riesgos de prompt injection para background agents](https://platform.openai.com/docs/codex/agent-network#risks-of-agent-internet-access).
5. Si Privacy Mode está deshabilitado, recopilamos prompts y entornos de desarrollo para mejorar el producto.
6. Si deshabilitas Privacy Mode al iniciar un background agent y luego lo habilitas durante la ejecución, el agente sigue con Privacy Mode deshabilitado hasta que termine.

<div id="dashboard-settings">
  ## Configuración del dashboard
</div>

Los admins del workspace pueden configurar opciones adicionales desde la pestaña Background Agents en el dashboard.

<div id="defaults-settings">
  ### Configuración predeterminada
</div>

* **Modelo predeterminado** – el modelo que se usa cuando una ejecución no especifica ninguno. Elige cualquier modelo compatible con Max Mode.
* **Repositorio predeterminado** – si está vacío, los agentes te piden que elijas un repo. Indicar un repo aquí te permite saltarte ese paso.
* **Rama base** – la rama desde la que los agentes crean un fork al abrir pull requests. Déjala en blanco para usar la rama predeterminada del repositorio.

<div id="security-settings">
  ### Configuración de seguridad
</div>

Todas las opciones de seguridad requieren privilegios de admin.

* **Restricciones de usuario** – elige *Ninguna* (todos los miembros pueden iniciar agentes en segundo plano) o *Lista de permitidos*. Cuando está en *Lista de permitidos*, especificas exactamente qué compas de equipo pueden crear agentes.
* **Seguimientos del equipo** – cuando está activado, cualquiera en el espacio de trabajo puede añadir mensajes de seguimiento a un agente que inició otra persona. Apágalo para restringir los seguimientos al propietario del agente y a los admins.
* **Mostrar resumen del agente** – controla si Cursor muestra las imágenes de diferencias de archivos del agente y los fragmentos de código. Desactívalo si prefieres no exponer rutas de archivo o código en la barra lateral.
* **Mostrar resumen del agente en canales externos** – extiende la opción anterior a Slack o cualquier canal externo que tengas conectado.

Los cambios se guardan al instante y se aplican de inmediato a los nuevos agentes.

# Agregar seguimiento
Source: https://docs.cursor.com/es/background-agent/api/add-followup

en/background-agent/api/openapi.yaml post /v0/agents/{id}/followup
Envía una instrucción adicional a un agente en segundo plano que está en ejecución.

# Conversación del agente
Source: https://docs.cursor.com/es/background-agent/api/agent-conversation

en/background-agent/api/openapi.yaml get /v0/agents/{id}/conversation
Recupera el historial de conversaciones de un agente en segundo plano.

Si el agente en segundo plano se eliminó, no vas a poder acceder a la conversación.

# Estado del agente
Source: https://docs.cursor.com/es/background-agent/api/agent-status

en/background-agent/api/openapi.yaml get /v0/agents/{id}
Obtén el estado actual y los resultados de un agente específico en segundo plano.

# Información de la clave de API
Source: https://docs.cursor.com/es/background-agent/api/api-key-info

en/background-agent/api/openapi.yaml get /v0/me
Obtén metadatos sobre la clave de API utilizada para la autenticación.

# Eliminar un agente
Source: https://docs.cursor.com/es/background-agent/api/delete-agent

en/background-agent/api/openapi.yaml delete /v0/agents/{id}
Eliminar permanentemente un agente en segundo plano y sus recursos asociados.

# Iniciar un agente
Source: https://docs.cursor.com/es/background-agent/api/launch-an-agent

en/background-agent/api/openapi.yaml post /v0/agents
Inicia un nuevo agente en segundo plano para trabajar en tu repositorio.

# Listar agentes
Source: https://docs.cursor.com/es/background-agent/api/list-agents

en/background-agent/api/openapi.yaml get /v0/agents
Obtén una lista paginada de todos los agentes en segundo plano del usuario autenticado.

# Listar modelos
Source: https://docs.cursor.com/es/background-agent/api/list-models

en/background-agent/api/openapi.yaml get /v0/models
Obtén una lista de modelos recomendados para agentes en segundo plano.

Si querés definir el modelo del agente en segundo plano al crearlo, podés usar este endpoint para ver una lista de modelos recomendados.

En ese caso, también te recomendamos ofrecer una opción "Auto", en la que no envíes un nombre de modelo al endpoint de creación y nosotros elegimos el modelo más adecuado.

# Listar repositorios de GitHub
Source: https://docs.cursor.com/es/background-agent/api/list-repositories

en/background-agent/api/openapi.yaml get /v0/repositories
Obtén una lista de repositorios de GitHub a los que tiene acceso el usuario autenticado.

<Warning>
  **Este endpoint tiene límites de tasa muy estrictos.**

Limita las solicitudes a **1/usuario/minuto** y **30/usuario/hora**.

Esta solicitud puede tardar decenas de segundos en responder para usuarios con acceso a muchos repositorios.

Asegúrate de manejar correctamente cuando esta información no esté disponible.
</Warning>

# Descripción general
Source: https://docs.cursor.com/es/background-agent/api/overview

Crea y administra por programación agentes en segundo plano que trabajen en tus repositorios

<div id="background-agents-api">
  # API de Background Agents
</div>

La API de Background Agents te permite crear y gestionar de forma programática agentes de codificación con IA que trabajan de manera autónoma en tus repositorios.
Podés usar la API para responder automáticamente al feedback de usuarios, corregir bugs, actualizar la documentación y mucho más.

<Info>
  La API de Background Agents está en beta, ¡nos encantaría tu feedback!
</Info>

<div id="key-features">
  ## Funcionalidades clave
</div>

* **Generación de código autónoma** - Crea agentes que entiendan tu prompt y hagan cambios en tu base de código
* **Integración con repositorios** - Trabaja directamente con repositorios de GitHub
* Prompts de seguimiento - Añade instrucciones adicionales a agentes en ejecución
* **Precios según uso** - Paga solo por los tokens que uses
* **Escalable** - Compatible con hasta 256 agentes activos por clave de API

<div id="quick-start">
  ## Inicio rápido
</div>

<div id="1-get-your-api-key">
  ### 1. Obtén tu clave de API
</div>

**Ve** a [Cursor Dashboard → Integrations](https://cursor.com/dashboard?tab=integrations) para crear tu clave de API.

<div id="2-start-using-the-api">
  ### 2. Empieza a usar la API
</div>

Todos los endpoints de la API son relativos a:

```
https://api.cursor.com
```

Consulta la [referencia de la API](/es/background-agent/api/launch-an-agent) para obtener una lista detallada de endpoints.

<div id="authentication">
  ## Autenticación
</div>

Todas las solicitudes a la API requieren autenticación con un token Bearer:

```
Authorization: Bearer TU_CLAVE_DE_API
```

Las claves de API se crean en el [Cursor Dashboard](https://cursor.com/dashboard?tab=integrations). Las claves están asociadas a tu cuenta y te permiten crear y administrar agentes (según los límites de tu plan y el acceso a los repositorios).

<div id="pricing">
  ## Precios
</div>

La API está actualmente en beta y tiene la misma tarifa que Background Agents. Los precios pueden cambiar a medida que escalemos el servicio. Consulta [Background Agent pricing](/es/account/pricing#background-agent).

<div id="next-steps">
  ## Próximos pasos
</div>

* Lee la [introducción general a Background Agents](/es/background-agent) para entender entornos, permisos y flujos de trabajo.
* Prueba Background Agents desde [web y móvil](/es/background-agent/web-and-mobile).
* Únete a la discusión en [Discord #background-agent](https://discord.gg/jfgpZtYpmb) o escribe a [background-agent-feedback@cursor.com](mailto:background-agent-feedback@cursor.com).

# Webhooks
Source: https://docs.cursor.com/es/background-agent/api/webhooks

Recibe notificaciones en tiempo real sobre cambios en el estado del agente en segundo plano

<div id="webhooks">
  # Webhooks
</div>

Cuando creas un agente con una URL de webhook, Cursor enviará solicitudes HTTP POST para notificarte sobre cambios de estado. Actualmente, solo se admiten eventos `statusChange`, específicamente cuando un agente entra en estado `ERROR` o `FINISHED`.

<div id="webhook-verification">
  ## Verificación de webhooks
</div>

Para asegurarte de que las solicitudes de webhook provienen auténticamente de Cursor, verificá la firma incluida con cada solicitud:

<div id="headers">
  ### Encabezados
</div>

Cada solicitud de webhook incluye los siguientes encabezados:

* **`X-Webhook-Signature`** – Contiene la firma HMAC-SHA256 con el formato `sha256=<hex_digest>`
* **`X-Webhook-ID`** – Un identificador único para esta entrega (útil para el logging)
* **`X-Webhook-Event`** – El tipo de evento (actualmente solo `statusChange`)
* **`User-Agent`** – Siempre establecido en `Cursor-Agent-Webhook/1.0`

<div id="signature-verification">
  ### Verificación de firma
</div>

Para verificar la firma del webhook, calculá la firma esperada y comparala con la firma recibida:

```javascript  theme={null}
const crypto = require('crypto');

function verifyWebhook(secret, rawBody, signature) {
  const firmaEsperada = 'sha256=' +
    crypto.createHmac('sha256', secret)
          .update(rawBody)
          .digest('hex');
  
  return signature === expectedSignature;
}
```

```python  theme={null}
import hmac
import hashlib

def verify_webhook(secret, raw_body, signature):
    firma_esperada = 'sha256=' + hmac.new(
        secret.encode(),
        raw_body,
        hashlib.sha256
    ).hexdigest()
    
    return signature == expected_signature
```

Usa siempre el cuerpo sin procesar de la solicitud (antes de cualquier parsing) al calcular la firma.

<div id="payload-format">
  ## Formato del payload
</div>

El payload del webhook se envía como JSON con la siguiente estructura:

```json  theme={null}
{
  "event": "statusChange",
  "timestamp": "2024-01-15T10:30:00Z",
  "id": "bc_abc123",
  "status": "FINISHED",
  "source": {
    "repository": "https://github.com/your-org/your-repo",
    "ref": "main"
  },
  "target": {
    "url": "https://cursor.com/agents?id=bc_abc123",
    "branchName": "cursor/add-readme-1234",
    "prUrl": "https://github.com/your-org/your-repo/pull/1234"
  },
  "summary": "Se agregó README.md con instrucciones de instalación"
}
```

Ten en cuenta que algunos campos son opcionales y solo se incluirán cuando estén disponibles.

<div id="best-practices">
  ## Mejores prácticas
</div>

* **Verifica las firmas** – Verifica siempre la firma del webhook para asegurarte de que la solicitud venga de Cursor
* **Maneja los reintentos** – Los webhooks pueden reintentarse si tu endpoint devuelve un código de estado de error
* **Responde rápido** – Devuelve un código de estado 2xx lo antes posible
* **Usa HTTPS** – Usa siempre URLs HTTPS para los endpoints de webhooks en producción
* **Almacena el payload sin procesar** – Guarda el payload del webhook sin procesar para depuración y verificación futuras

# Web y móvil
Source: https://docs.cursor.com/es/background-agent/web-and-mobile

Ejecuta agentes de código desde cualquier dispositivo con traspaso fluido al escritorio

<div id="overview">
  ## Descripción general
</div>

El agente de Cursor en la web lleva un potente asistente de código a cualquier dispositivo. Ya sea que estés con el teléfono mientras caminás o trabajando en el navegador, ahora podés arrancar agentes de código potentes que trabajan en segundo plano.
Cuando terminen, retomá su trabajo en Cursor, revisá y mergeá cambios, o compartí enlaces con tu equipo para colaborar.

Empezá en [cursor.com/agents](https://cursor.com/agents).

<div id="getting-started">
  ## Primeros pasos
</div>

<div id="quick-setup">
  ### Configuración rápida
</div>

1. **Visita la app web**: Entra a [cursor.com/agents](https://cursor.com/agents) desde cualquier dispositivo
2. **Inicia sesión**: Accede con tu cuenta de Cursor
3. **Conecta GitHub**: Vincula tu cuenta de GitHub para acceder a los repositorios
4. **Inicia tu primer agente**: Escribe una tarea y mira cómo el agente se pone a trabajar

<div id="mobile-installation">
  ### Instalación en móvil
</div>

Para la mejor experiencia en móvil, instala Cursor como una app web progresiva (PWA):

* **iOS**: Abre [cursor.com/agents](https://cursor.com/agents) en Safari, toca el botón de compartir y luego “Add to Home Screen”
* **Android**: Abre la URL en Chrome, toca el menú y luego “Add to Home Screen” o “Install App”

<Tip>
  Instalar como PWA ofrece una experiencia similar a una app nativa con: - Interfaz de pantalla completa - Arranques más rápidos - Ícono de la app en tu pantalla de inicio
</Tip>

<div id="working-across-devices">
  ## Trabajar en varios dispositivos
</div>

El Web and Mobile Agent está diseñado para funcionar con tu flujo de trabajo en el escritorio; haz clic en "Open in Cursor" para continuar el trabajo del agente en tu IDE.

<div id="team-collaboration">
  ### Colaboración en equipo
</div>

* **Acceso compartido**: Comparte enlaces con miembros del equipo para colaborar en ejecuciones del agente.
* **Proceso de revisión**: Los colaboradores pueden revisar diffs y dar feedback.
* **Gestión de pull requests**: Crea, revisa y fusiona pull requests directamente desde la interfaz web.

<div id="slack-integration">
  ### Integración con Slack
</div>

Lanza agentes directamente desde Slack mencionando `@Cursor`, y, al iniciar agentes desde la web o el móvil, elige recibir notificaciones de Slack al finalizar.

<Card title="Usar Cursor en Slack" icon="slack" href="/es/slack">
  Obtén más información sobre cómo configurar y usar la integración con Slack, incluyendo
  cómo disparar agentes y recibir notificaciones.
</Card>

<div id="pricing">
  ## Precios
</div>

Los agentes web y móviles usan el mismo modelo de precios que los Agentes en Segundo Plano.

Obtén más información sobre los [precios de los Agentes en Segundo Plano](/es/account/pricing#background-agent).

<div id="troubleshooting">
  ## Solución de problemas
</div>

<AccordionGroup>
  <Accordion title="Agent runs are not starting">
    * Asegúrate de haber iniciado sesión y de tener conectada tu cuenta de GitHub. - Revisa
      que tengas los permisos necesarios en el repositorio. - También necesitas
      estar en una Prueba Pro o en un plan de pago con la facturación por uso habilitada. Para habilitar
      la facturación por uso, ve a la pestaña de configuración de tu
      [Dashboard](https://www.cursor.com/dashboard?tab=settings).
  </Accordion>

<Accordion title="Can't see agent runs on mobile">
    Intenta actualizar la página o borrar la caché de tu navegador. Asegúrate de usar
    la misma cuenta en todos tus dispositivos.
  </Accordion>

<Accordion title="Slack integration not working">
    Verifica que el administrador de tu espacio de trabajo haya instalado la app de Slack de Cursor y que
    tengas los permisos necesarios.
  </Accordion>
</AccordionGroup>

# Bugbot
Source: https://docs.cursor.com/es/bugbot

Revisión de código con IA para pull requests

Bugbot revisa pull requests e identifica bugs, problemas de seguridad y de calidad de código.

<Tip>
  Bugbot incluye un plan gratuito: cada usuario recibe una cantidad limitada de revisiones de PR gratis cada mes. Cuando alcanzas el límite, las revisiones se pausan hasta tu próximo ciclo de facturación. Podés actualizar en cualquier momento a una prueba Pro gratuita de 14 días para revisiones ilimitadas (sujeto a controles estándar contra abusos).
</Tip>

<div id="how-it-works">
  ## Cómo funciona
</div>

Bugbot analiza los diffs de los PR y deja comentarios con explicaciones y sugerencias de corrección. Se ejecuta automáticamente en cada actualización del PR o manualmente cuando se activa.

* Ejecuta **revisiones automáticas** en cada actualización del PR
* **Disparo manual** comentando `cursor review` o `bugbot run` en cualquier PR
* Los enlaces **Fix in Cursor** abren issues directamente en Cursor
* Los enlaces **Fix in Web** abren issues directamente en [cursor.com/agents](https://cursor.com/agents)

<div id="setup">
  ## Configuración
</div>

Requiere acceso de admin en Cursor y acceso de admin en la organización de GitHub.

1. Ve a [cursor.com/dashboard](https://cursor.com/dashboard?tab=bugbot)
2. Entra a la pestaña Bugbot
3. Haz clic en `Connect GitHub` (o en `Manage Connections` si ya está conectado)
4. Sigue el flujo de instalación de GitHub
5. Vuelve al dashboard para habilitar Bugbot en repositorios específicos

<div id="setup">
  ## Configuración
</div>

<Tabs defaultValue="Team">
  <Tab title="Individual">
    ### Configuración del repositorio

Activa o desactiva Bugbot por repositorio desde tu lista de instalaciones. Bugbot solo se ejecuta en PR que tú crees.

### Configuración personal

* Ejecutar **solo cuando se le mencione** comentando `cursor review` o `bugbot run`
    * Ejecutar **solo una vez** por PR, omitiendo commits posteriores
  </Tab>

<Tab title="Team">
    ### Configuración del repositorio

Los administradores del equipo pueden activar Bugbot por repositorio, configurar listas de permitidos/denegados para revisores y definir:

* Ejecutar **solo una vez** por PR y por instalación, omitiendo commits posteriores
    * **Desactivar las revisiones en línea** para evitar que Bugbot deje comentarios directamente en las líneas de código

Bugbot se ejecuta para todas las personas que contribuyan en los repositorios habilitados, sin importar la pertenencia al equipo.

### Configuración personal

Los miembros del equipo pueden anular la configuración para sus propios PR:

* Ejecutar **solo cuando se le mencione** comentando `cursor review` o `bugbot run`
    * Ejecutar **solo una vez** por PR, omitiendo commits posteriores
    * **Habilitar revisiones en PR en borrador** para incluir pull requests en borrador en las revisiones automáticas
  </Tab>
</Tabs>

<div id="analytics">
  ### Analítica
</div>

<div id="rules">
  ## Reglas
</div>

Crea archivos `.cursor/BUGBOT.md` para aportar contexto específico del proyecto en las revisiones. Bugbot siempre incluye el archivo raíz `.cursor/BUGBOT.md` y cualquier archivo adicional que encuentre al recorrer hacia arriba desde los archivos modificados.

```
project/
  .cursor/BUGBOT.md          # Siempre incluido (reglas a nivel de proyecto)
  backend/
    .cursor/BUGBOT.md        # Incluido al revisar archivos del backend
    api/
      .cursor/BUGBOT.md      # Incluido al revisar archivos del API
  frontend/
    .cursor/BUGBOT.md        # Incluido al revisar archivos del frontend
```

<AccordionGroup>
  <Accordion title="Ejemplo .cursor/BUGBOT.md">
    ```markdown  theme={null}
    # Guías para la revisión del proyecto

## Enfoques de seguridad

- Validar la entrada del usuario en los endpoints de la API
    - Revisar vulnerabilidades de inyección SQL en las consultas a la base de datos
    - Garantizar una autenticación adecuada en las rutas protegidas

## Patrones de arquitectura

- Usar inyección de dependencias para los servicios
    - Seguir el patrón de repositorio para el acceso a datos
    - Implementar un manejo de errores adecuado con clases de error personalizadas

## Problemas comunes

- Fugas de memoria en componentes de React (revisa la limpieza de useEffect)
    - Falta de límites de error (error boundaries) en componentes de UI
    - Convenciones de nombres inconsistentes (usa camelCase para las funciones)

```
  </Accordion>
</AccordionGroup>

<div id="pricing">
  ## Precios
</div>

Bugbot ofrece dos planes: **Gratis** y **Pro**.

Cada usuario obtiene una cantidad limitada de revisiones de PR gratis cada mes. En equipos, cada miembro del equipo recibe sus propias revisiones gratis. Cuando llegás al límite, las revisiones se pausan hasta tu próximo ciclo de facturación. Podés pasar en cualquier momento a la prueba Pro gratuita de 14 días para tener revisiones ilimitadas.

<Tabs defaultValue="Teams">
  <Tab title="Individuals">
    ### Tarifa fija

USD 40 al mes por revisiones ilimitadas de Bugbot en hasta 200 PR al mes en todos los repositorios.

### Primeros pasos

Suscribite desde la configuración de tu cuenta.
  </Tab>

<Tab title="Teams">
    ### Facturación por usuario

Los equipos pagan USD 40 por usuario por mes por revisiones ilimitadas.

Consideramos usuario a quien haya sido autor de PR revisados por Bugbot en un mes.

Todas las licencias se liberan al inicio de cada ciclo de facturación y se asignan por orden de llegada. Si un usuario no autoriza ningún PR revisado por Bugbot en un mes, el cupo puede ser usado por otra persona.

### Límites de cupos

Los admins del equipo pueden establecer un máximo de cupos de Bugbot por mes para controlar los costos.

### Primeros pasos

Suscribite desde el panel de tu equipo para habilitar la facturación.

### Medidas contra abusos

Para evitar abusos, tenemos un tope conjunto de 200 pull requests por mes por cada licencia de Bugbot. Si necesitás más de 200 pull requests por mes, contactanos en [hi@cursor.com](mailto:hi@cursor.com) y con gusto te ayudamos.

Por ejemplo, si tu equipo tiene 100 usuarios, tu organización inicialmente podrá revisar 20.000 pull requests por mes. Si alcanzan ese límite de forma natural, por favor escribinos y con gusto aumentamos el límite.
  </Tab>
</Tabs>

<div id="troubleshooting">
  ## Solución de problemas
</div>

Si Bugbot no está funcionando:

1. **Activa el modo detallado** añadiendo un comentario `cursor review verbose=true` o `bugbot run verbose=true` para obtener logs detallados y el ID de la solicitud
2. **Revisa los permisos** para verificar que Bugbot tenga acceso al repositorio
3. **Verifica la instalación** para confirmar que la app de GitHub esté instalada y habilitada

Incluye el ID de la solicitud del modo detallado al reportar problemas.

<div id="faq">
  ## Preguntas frecuentes
</div>

<AccordionGroup>
  <Accordion title="¿Bugbot es compatible con el modo de privacidad?">
    Sí, Bugbot cumple las mismas normas de privacidad que Cursor y procesa los datos de la misma forma que otras solicitudes de Cursor.
  </Accordion>

<Accordion title="¿Qué pasa cuando alcanzó el límite del plan gratuito?">
    Cuando llegas a tu límite mensual del plan gratuito, las revisiones de Bugbot se pausan hasta tu siguiente ciclo de facturación. Puedes pasar a la prueba Pro gratuita de 14 días para obtener revisiones ilimitadas (sujetas a las protecciones estándar contra abuso).
  </Accordion>
</AccordionGroup>

```
```

# Code Review
Source: https://docs.cursor.com/es/cli/cookbook/code-review

Crea un flujo de trabajo de GitHub Actions que use Cursor CLI para revisar automáticamente los pull requests y ofrecer comentarios

Este tutorial te muestra cómo configurar el code review usando Cursor CLI en GitHub Actions. El flujo de trabajo analizará los pull requests, identificará problemas y publicará comentarios como respuestas.

<Tip>
  Para la mayoría de usuarios, te recomendamos usar [Bugbot](/es/bugbot). Bugbot ofrece code review automatizado y gestionado sin necesidad de configuración. Este enfoque con la CLI es útil para explorar capacidades y para personalización avanzada.
</Tip>

<div className="space-y-4">
  <Expandable title="archivo completo del flujo de trabajo">
    ```yaml cursor-code-review.yml theme={null}
    name: Code Review

on:
      pull_request:
        types: [opened, synchronize, reopened, ready_for_review]

permissions:
      pull-requests: write
      contents: read
      issues: write

jobs:
      code-review:
        runs-on: ubuntu-latest
        # Omitir la revisión de código automatizada para PR en borrador
        if: github.event.pull_request.draft == false
        steps:
          - name: Checkout repository
            uses: actions/checkout@v4
            with:
              fetch-depth: 0
              ref: ${{ github.event.pull_request.head.sha }}

- name: Install Cursor CLI
            run: |
              curl https://cursor.com/install -fsS | bash
              echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
            run: |
              git config user.name "Cursor Agent"
              git config user.email "cursoragent@cursor.com"

- name: Perform automated code review
            env:
              CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
              MODEL: gpt-5
              GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
              BLOCKING_REVIEW: ${{ vars.BLOCKING_REVIEW || 'false' }}
            run: |
              cursor-agent --force --model "$MODEL" --output-format=text --print 'Estás operando en un runner de GitHub Actions realizando una revisión de código automatizada. La CLI de gh está disponible y autenticada mediante GH_TOKEN. Puedes comentar en pull requests.

Contexto:
              - Repo: ${{ github.repository }}
              - PR Number: ${{ github.event.pull_request.number }}
              - PR Head SHA: ${{ github.event.pull_request.head.sha }}
              - PR Base SHA: ${{ github.event.pull_request.base.sha }}
              - Blocking Review: ${{ env.BLOCKING_REVIEW }}

Objetivos:
              1) Volver a revisar los comentarios existentes y responder resuelto cuando ya estén atendidos.
              2) Revisar el diff actual del PR y señalar solo problemas claros y de alta severidad.
              3) Dejar comentarios en línea muy breves (1‑2 frases) solo en las líneas modificadas y un breve resumen al final.

Procedimiento:
              - Get existing comments: gh pr view --json comments
              - Get diff: gh pr diff
              - Get changed files with patches to compute inline positions: gh api repos/${{ github.repository }}/pulls/${{ github.event.pull_request.number }}/files --paginate --jq '.[] | {filename,patch}'
              - Calcular anclajes exactos en línea para cada problema (ruta de archivo + posición en el diff). Los comentarios DEBEN colocarse en línea en la línea modificada del diff, no como comentarios de nivel superior.
              - Detectar comentarios previos de nivel superior del tipo "sin problemas" escritos por este bot (coincidir cuerpos como: "✅ no issues", "No issues found", "LGTM").
              - Si la ejecución ACTUAL encuentra problemas y existe cualquier comentario previo de "sin problemas":
                - Preferir eliminarlos para evitar confusión:
                  - Try deleting top-level issue comments via: gh api -X DELETE repos/${{ github.repository }}/issues/comments/<comment_id>
                  - Si no es posible eliminarlos, minimízalos vía GraphQL (minimizeComment) o edítalos para anteponer "[Superseded by new findings]".
                - Si no es posible ni eliminar ni minimizar, responde a ese comentario: "⚠️ Superseded: issues were found in newer commits".
              - Si un problema reportado previamente parece estar corregido por cambios cercanos, responde: ✅ Este problema parece estar resuelto por los cambios recientes
              - Analizar SOLO:
                - Desreferencias de null/undefined
                - Fugas de recursos (archivos o conexiones sin cerrar)
                - Inyección (SQL/XSS)
                - Concurrencia/condiciones de carrera
                - Falta de manejo de errores en operaciones críticas
                - Errores lógicos evidentes con comportamiento incorrecto
                - Patrones claros anti‑rendimiento con impacto medible
                - Vulnerabilidades de seguridad definitivas
              - Evitar duplicados: omite si ya existe feedback similar en o cerca de las mismas líneas.

Reglas para comentar:
              - Máximo 10 comentarios en línea en total; prioriza los problemas más críticos
              - Un problema por comentario; colócalo exactamente en la línea modificada
              - Todos los comentarios de problemas DEBEN ser en línea (anclados a un archivo y línea/posición en el diff del PR)
              - Tono natural, específico y accionable; no menciones que es automatizado ni el nivel de confianza
              - Usa emojis: 🚨 Crítico 🔒 Seguridad ⚡ Rendimiento ⚠️ Lógica ✅ Resuelto ✨ Mejora

Envío:
              - Si NO hay problemas que reportar y ya existe un comentario de nivel superior indicando "sin problemas" (p. ej., "✅ no issues", "No issues found", "LGTM"), NO envíes otro comentario. Omite el envío para evitar redundancia.
              - Si NO hay problemas que reportar y NO existe un comentario previo de "sin problemas", envía un comentario de resumen breve indicando que no hay problemas.
              - Si HAY problemas que reportar y existe un comentario previo de "sin problemas", asegúrate de eliminar/minimizar/marcar como reemplazado el comentario previo antes de enviar la nueva revisión.
              - Si HAY problemas que reportar, envía UNA revisión que contenga SOLO comentarios en línea más un resumen conciso opcional. Usa la API de GitHub Reviews para asegurar que los comentarios sean en línea:
                - Build a JSON array of comments like: [{ "path": "<file>", "position": <diff_position>, "body": "..." }]
                - Submit via: gh api repos/${{ github.repository }}/pulls/${{ github.event.pull_request.number }}/reviews -f event=COMMENT -f body="$SUMMARY" -f comments='[$COMMENTS_JSON]'
              - NO uses: gh pr review --approve o --request-changes

Comportamiento de bloqueo:
              - If BLOCKING_REVIEW is true and any 🚨 or 🔒 issues were posted: echo "CRITICAL_ISSUES_FOUND=true" >> $GITHUB_ENV
              - Otherwise: echo "CRITICAL_ISSUES_FOUND=false" >> $GITHUB_ENV
              - Always set CRITICAL_ISSUES_FOUND at the end
              '

- name: Check blocking review results
            if: env.BLOCKING_REVIEW == 'true'
            run: |
              echo "Comprobando problemas críticos..."
              echo "CRITICAL_ISSUES_FOUND: ${CRITICAL_ISSUES_FOUND:-unset}"

if [ "${CRITICAL_ISSUES_FOUND:-false}" = "true" ]; then
                echo "❌ Se encontraron problemas críticos y la revisión bloqueante está habilitada. Fallando el flujo de trabajo."
                exit 1
              else
                echo "✅ No se encontraron problemas bloqueantes."
              fi
    ```
  </Expandable>

<div id="configure-authentication">
  ## Configura la autenticación
</div>

[Configura tu API key y los secretos del repositorio](/es/cli/github-actions#authentication) para autenticar Cursor CLI en GitHub Actions.

<div id="set-up-agent-permissions">
  ## Configura los permisos del agente
</div>

Crea un archivo de configuración para controlar qué acciones puede realizar el agente. Esto evita operaciones no deseadas, como hacer push de código o crear pull requests.

Crea `.cursor/cli.json` en la raíz de tu repositorio:

```json  theme={null}
{
  "permissions": {
    "deny": [
      "Shell(git push)",
      "Shell(gh pr create)",
      "Escribir(**)"
    ]
  }
}
```

Esta configuración permite que el agente lea archivos y use la CLI de GitHub para comentarios, pero impide que haga cambios en tu repositorio. Consulta la [referencia de permisos](/es/cli/reference/permissions) para ver más opciones de configuración.

<div id="build-the-github-actions-workflow">
  ## Crea el flujo de trabajo de GitHub Actions
</div>

Ahora vamos a armar el flujo de trabajo paso a paso.

<div id="set-up-the-workflow-trigger">
  ### Configura el disparador del flujo de trabajo
</div>

Crea `.github/workflows/cursor-code-review.yml` y configúralo para ejecutarse en pull requests:

```yaml  theme={null}
name: Revisión de código de Cursor

on:
  pull_request:
    types: [opened, synchronize, reopened, ready_for_review]

jobs:
  code-review:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write
    
    steps:
```

<div id="checkout-the-repository">
  ### Revisa el repositorio
</div>

Agrega el paso de checkout para acceder al código del pull request:

```yaml  theme={null}
- name: Checkout del repositorio
  uses: actions/checkout@v4
  with:
    fetch-depth: 0
    ref: ${{ github.event.pull_request.head.sha }}
```

<div id="install-cursor-cli">
  ### Instala la CLI de Cursor
</div>

Agrega el paso de instalación de la CLI:

```yaml  theme={null}
- name: Instalar Cursor CLI
  run: |
    curl https://cursor.com/install -fsS | bash
    echo "$HOME/.cursor/bin" >> $GITHUB_PATH
```

<div id="configure-the-review-agent">
  ### Configura el agente de revisión
</div>

Antes de implementar el paso de revisión completo, entendamos la anatomía de nuestro prompt de revisión. Esta sección describe cómo queremos que se comporte el agente:

**Objetivo**:
Queremos que el agente revise el diff del PR actual y marque solo problemas claros y de alta severidad, luego deje comentarios en línea muy cortos (1-2 frases) solo en líneas modificadas, con un breve resumen al final. Esto mantiene una buena relación señal-ruido.

**Formato**:
Queremos comentarios breves y directos. Usamos emojis para facilitar el escaneo de los comentarios y queremos un resumen de alto nivel de la revisión completa al final.

**Envío**:
Cuando termine la revisión, queremos que el agente incluya un comentario corto basado en lo encontrado durante la revisión. El agente debe enviar una sola revisión que contenga comentarios en línea más un resumen conciso.

**Casos límite**:
Necesitamos manejar:

* Comentarios existentes resueltos: el agente debe marcarlos como listos cuando se aborden
* Evitar duplicados: el agente debe omitir comentar si ya existe feedback similar en o cerca de las mismas líneas

**Prompt final**:
El prompt completo combina todos estos requisitos de comportamiento para crear feedback enfocado y accionable

Ahora implementemos el paso del agente de revisión:

```yaml  theme={null}
- name: Realizar revisión de código
  env:
    CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
    GH_TOKEN: ${{ github.token }}
  run: |
    cursor-agent --force --model "$MODEL" --output-format=text --print "Estás operando en un runner de GitHub Actions realizando una revisión de código automatizada. La CLI de gh está disponible y autenticada mediante GH_TOKEN. Podés comentar en pull requests.
    
    Contexto:
    - Repo: ${{ github.repository }}
    - PR Number: ${{ github.event.pull_request.number }}
    - PR Head SHA: ${{ github.event.pull_request.head.sha }}
    - PR Base SHA: ${{ github.event.pull_request.base.sha }}
    
    Objetivos:
    1) Volver a revisar los comentarios existentes y responder resuelto cuando estén atendidos
    2) Revisar el diff actual del PR y señalar solo problemas claros y de alta severidad
    3) Dejar comentarios en línea muy cortos (1-2 frases) solo en líneas modificadas y un resumen breve al final
    
    Procedimiento:
    - Obtener comentarios existentes: gh pr view --json comments
    - Obtener diff: gh pr diff
    - Si un problema reportado anteriormente parece solucionado por cambios cercanos, responder: ✅ Este problema parece estar resuelto por los cambios recientes
    - Evitar duplicados: omitir si ya existe feedback similar en o cerca de las mismas líneas
    
    Reglas para comentar:
    - Máximo 10 comentarios en línea en total; priorizar los problemas más críticos
    - Un problema por comentario; colocarlo en la línea exacta modificada
    - Tono natural, específico y accionable; no mencionar automatización ni alta confianza
    - Usar emojis: 🚨 Crítico 🔒 Seguridad ⚡ Rendimiento ⚠️ Lógica ✅ Resuelto ✨ Mejora
    
    Envío:
    - Enviar una sola revisión que contenga comentarios en línea más un resumen conciso
    - Usar solo: gh pr review --comment
    - No usar: gh pr review --approve o --request-changes"
```

```text  theme={null}
.
├── .cursor/
│   └── cli.json
├── .github/
│   └── workflows/
│       └── cursor-code-review.yml
```

<div id="test-your-reviewer">
  ## Prueba tu reviewer
</div>

Crea un pull request de prueba para verificar que el flujo de trabajo funcione y que el agente publique comentarios de revisión con emojis y feedback.

<div id="next-steps">
  ## Próximos pasos
</div>

Ahora tienes un sistema de revisión de código automatizada en marcha. Considera estas mejoras:

* Configura flujos de trabajo adicionales para [corregir fallos de CI](/es/cli/cookbook/fix-ci)
* Configura distintos niveles de revisión para diferentes ramas
* Intégralo con el proceso de revisión de código existente de tu equipo
* Personaliza el comportamiento del agente para distintos tipos de archivos o directorios

<Expandable title="Avanzado: Revisiones bloqueantes">
  Puedes configurar el flujo de trabajo para que falle si se encuentran problemas críticos, evitando que se fusione el pull request hasta que se aborden.

**Añadir comportamiento bloqueante al prompt**

Primero, actualiza el paso de tu agente de revisión para incluir la variable de entorno `BLOCKING_REVIEW` y añade este comportamiento bloqueante al prompt:

```
  Comportamiento bloqueante:
  - Si BLOCKING_REVIEW es true y se publicaron problemas 🚨 o 🔒: echo "CRITICAL_ISSUES_FOUND=true" >> $GITHUB_ENV
  - De lo contrario: echo "CRITICAL_ISSUES_FOUND=false" >> $GITHUB_ENV
  - Establece siempre CRITICAL_ISSUES_FOUND al final
  ```

**Añadir el paso de verificación bloqueante**

Luego, añade este nuevo paso después del paso de revisión de código:

if [ "${CRITICAL_ISSUES_FOUND:-false}" = "true" ]; then
              echo "❌ Critical issues found and blocking review is enabled. Failing the workflow."
              exit 1
            else
              echo "✅ No blocking issues found."
            fi
  ```
</Expandable>

# Corregir fallos de CI
Source: https://docs.cursor.com/es/cli/cookbook/fix-ci

Corrige problemas de CI en un repositorio usando Cursor CLI en GitHub Actions

Corrige fallos de CI usando Cursor CLI en GitHub Actions. Este workflow analiza los fallos, aplica correcciones puntuales y crea una rama de fix con un enlace para crear un PR al instante.

Este workflow supervisa por nombre un workflow concreto. Actualiza la lista `workflows` para que coincida con el nombre real de tu workflow de CI.

<CodeGroup>
  ```yaml auto-fix-ci.yml theme={null}
  name: Fix CI Failures

on:
    workflow_run:
      workflows: [Test]
      types: [completed]

permissions:
    contents: write
    pull-requests: write
    actions: read

- name: Install Cursor CLI
          run: |
            curl https://cursor.com/install -fsS | bash
            echo "$HOME/.cursor/bin" >> $GITHUB_PATH

- name: Configure git identity
          run: |
            git config user.name "Cursor Agent"
            git config user.email "cursoragent@cursor.com"

# Goal:
            - Implement an end-to-end CI fix flow driven by the failing PR, creating a separate persistent fix branch and proposing a quick-create PR back into the original PR's branch.

```
</CodeGroup>

# Auditoría de secretos
Source: https://docs.cursor.com/es/cli/cookbook/secret-audit

Audita los secretos de un repositorio usando Cursor CLI en GitHub Actions

Audita tu repo para detectar vulnerabilidades de seguridad y exposición de secretos usando Cursor CLI. Este workflow busca posibles secretos, detecta patrones de workflow riesgosos y propone correcciones de seguridad.