# FastAPI-MCP — Turn Any FastAPI into MCP Tools

> Zero-config bridge that exposes existing FastAPI endpoints as MCP tools with authentication. Add one line to make your API AI-accessible.

## Install

Merge the JSON below into your `.mcp.json`:

# FastAPI-MCP — Turn Any FastAPI into MCP Tools

## Quick Use

```bash
pip install fastapi-mcp
```

```python
from fastapi import FastAPI
from fastapi_mcp import FastApiMCP

app = FastAPI()

@app.get("/items/{item_id}")
async def get_item(item_id: int):
    return {"item_id": item_id, "name": "Example"}

# One line to expose all endpoints as MCP tools
mcp = FastApiMCP(app)
mcp.mount()
```

Your FastAPI app now serves MCP at `/mcp` — connect any AI assistant to it.

## Introduction

FastAPI-MCP is a Python library that **automatically converts your existing FastAPI endpoints into MCP-compatible tools**. Instead of writing MCP server code from scratch, you add a single line to your existing FastAPI app and every endpoint becomes an AI-callable tool.

Key features:

- **Zero Config** — `FastApiMCP(app).mount()` is all you need. Every route becomes an MCP tool automatically.
- **Authentication Built-in** — Leverages your existing FastAPI security dependencies (OAuth2, API keys, etc.) for MCP tool access.
- **Schema Preservation** — Request/response models, validation rules, and documentation carry over to MCP tool definitions.
- **Selective Exposure** — Include or exclude specific endpoints with `include_operations` and `exclude_operations`.
- **ASGI Transport** — Direct in-process communication eliminates HTTP overhead when the MCP client and FastAPI run together.
- **Custom Tool Naming** — Override auto-generated tool names for cleaner AI assistant integration.

Perfect for teams that already have FastAPI backends and want to make them AI-accessible without rewriting anything.

## FAQ

**Q: Does it work with existing authentication?**
A: Yes. FastAPI-MCP reuses your existing FastAPI Depends() security — OAuth2 bearer tokens, API keys, and custom auth all work transparently.

**Q: Can I choose which endpoints to expose?**
A: Yes. Use `include_operations=["get_item", "list_items"]` or `exclude_operations=["delete_all"]` to control access.

**Q: How do I connect Claude to my FastAPI-MCP server?**
A: Add this to your MCP config:
```json
{
  "mcpServers": {
    "my-api": {
      "url": "http://localhost:8000/mcp"
    }
  }
}
```

**Q: Does it handle complex request bodies?**
A: Yes. Pydantic models, query parameters, path parameters, and JSON bodies are all converted to MCP tool input schemas automatically.

## Works With

- FastAPI 0.100+
- Python 3.10+
- Claude Code / Claude Desktop / Any MCP client
- Any existing FastAPI authentication setup

## Source & Thanks

- GitHub: [tadata-org/fastapi_mcp](https://github.com/tadata-org/fastapi_mcp)
- License: MIT
- Stars: 11,000+
- Maintainer: Tadata

Thanks to the Tadata team for the elegant "one line to MCP" approach that saves backend teams weeks of integration work.


---
Source: https://tokrepo.com/en/workflows/b764314c-4294-449c-9bb8-c8b1f70098a3
Author: MCP Hub