fastapi_mcp

FastAPI-MCP is a Python library that takes an existing FastAPI web application and automatically exposes all of its API endpoints as tools that AI assistants can call. FastAPI is a popular Python framework for building web APIs. MCP, the Model Context Protocol, is a standard that lets AI tools like Claude or Cursor discover and invoke external capabilities during a conversation.

The idea is that if you already have a working FastAPI backend, you should not need to rebuild anything to make it usable by an AI assistant. You add this library, call two functions, and your entire API becomes available to AI tools at a new URL. The library reads your existing FastAPI route definitions, including their documentation and data schemas, and uses that information to describe each endpoint to the AI in the format MCP requires.

The README highlights that this is built as a native FastAPI extension rather than a generic converter. That distinction matters for authentication: instead of creating a separate authentication layer for the MCP interface, you can reuse the same FastAPI dependency functions you already wrote for your regular API. The library also communicates with your FastAPI app directly through its internal interface rather than making HTTP calls to it, which the README notes is more efficient.

Deployment is flexible. You can mount the MCP server onto the same application so it runs as part of your existing service, or you can run it as a separate process if your infrastructure requires that.

Installation is through pip or the uv package manager. Python 3.10 or later is required, with 3.12 recommended. The project is open-source under the MIT license and is maintained by Tadata, which also offers a managed hosted version of the same functionality for teams that prefer not to self-host.