> ## Documentation Index
> Fetch the complete documentation index at: https://docs.koalr.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI agents (MCP)

> Connect Claude, Codex, Cursor or any MCP-capable agent to read your Koalr data.

<video autoPlay muted loop playsInline className="w-full rounded-xl border border-gray-200/60 dark:border-gray-800" src="https://mintcdn.com/koalr-bf7c8d6b/y1QNqY4huQd8xIv6/images/clips/integrations-mcp.mp4?fit=max&auto=format&n=y1QNqY4huQd8xIv6&q=85&s=972af2ce7d65fb67aeb576171d0eeca5" poster="/images/clips/integrations-mcp.jpg" aria-label="The MCP setup guides and tool list" data-path="images/clips/integrations-mcp.mp4" />

**Integrations, AI agents (MCP)** is the in-app home for Koalr's MCP server: setup guides with your server URL pre-filled, the list of tools, and what your agent can and cannot reach.

Full documentation is in the API tab:

<CardGroup cols={2}>
  <Card title="Introduction" icon="book-open" href="/api/mcp/introduction">
    What the server is, what it can do, and the principles it follows.
  </Card>

  <Card title="Setup" icon="plug" href="/api/mcp/setup">
    Claude Code, Claude.ai, Codex, Cursor and other clients.
  </Card>

  <Card title="Tools reference" icon="wrench" href="/api/mcp/tools">
    All fifteen tools with their arguments.
  </Card>

  <Card title="Limits and security" icon="shield" href="/api/mcp/limits-and-security">
    Rate limits, read-only guarantee, permission scoping.
  </Card>
</CardGroup>

## In short

* Server URL: `https://app.koalr.ai/api/mcp`
* Sign in with your own Koalr account; a consent screen shows what is granted.
* Read-only, always. Your permissions, nothing more.
* Remove the server from your client to disconnect. Sign-ins expire within 24 hours, and membership is checked live on every call.