FastMCP - FastAPI MCP Framework & Runner MCP server

fast-mcp

<!-- mcp-name: io.github.Manas-maker/fast-mcp -->

FastAPI-native Model Context Protocol (MCP) framework with automatic route reflection, ASGI scope bridging, dynamic progressive tool discovery, resilient error recovery, and interactive in-chat MCP Apps (SEP-1865).

Highlights

• ⚡ Hybrid Dual-Citizen ASGI Mount: Mounts directly onto any existing FastAPI instance in-process over standard ASGI—zero external proxying, zero hanging subprocesses.
• 🔌 Local stdio CLI Runner & Desktop AI Bridge: Run fast-mcp stdio main:app or python -m fast_mcp stdio main:app to connect desktop AI clients (Claude Desktop, Cursor) over standard I/O pipes with zero network setup.
• 🔍 Route Reflection: Opt-in tags (tags=["mcp"]) automatically convert FastAPI endpoints, Pydantic models, docstrings, path/query/body parameters into MCP tools.
• 🛠️ Custom AI Tools (@mcp.tool): Define AI-tailored composite tools alongside reflected routes with automatic schema and docstring extraction.
• 🔐 ASGI Scope Bridging: Client authorization headers (Authorization: Bearer , cookies, API keys) captured during the MCP handshake are bridged into an in-memory ASGI Request, natively resolving FastAPI's Depends() and Security() providers without code changes.
• 🧠 Dynamic Progressive Tool Discovery: Protect agent context windows via progressive discovery (dynamicdiscovery=True), the searchtools(query: str) meta-tool, and zero-dependency KeywordTagRouter.
• 🛡️ Resilient Error Recovery & Minified JSON: Traps route HTTPException and Pydantic validation errors into informative CallToolResult(isError=True) responses so LLMs can self-correct without protocol failures. Output defaults to compact, token-conscious minified JSON with custom @mcp.serializer formatting hooks.
• 🖥️ Dual UI & In-Chat MCP Apps (SEP-1865): Embedded browser inspector at /mcp/docs plus native support for in-chat interactive iframes in desktop AI clients (Claude Desktop, Cursor, VS Code) via _meta.ui.resourceUri, the built-in inspect() tool, and the @mcp.app() decorator.

Installation

pip install mcp-fastapi

Or using uv:

uv add mcp-fastapi

Quickstart

from fastapi import FastAPI, Depends, Header, HTTPException
from pydantic import BaseModel, Field
from fast_mcp import FastMCP

app = FastAPI(title="Store API")
mcp = FastMCP(app=app, name="store-mcp")

# 1. Existing FastAPI route reflected automatically via tags=["mcp"]
class Product(BaseModel):
    id: int
    name: str
    price: float

@app.get("/products/{product_id}", tags=["mcp"])
async def get_product(product_id: int) -> Product:
    """Fetch product details by ID."""
    if product_id == 404:
        raise HTTPException(status_code=404, detail="Product not found")
    return Product(id=product_id, name="Smart Widget", price=29.99)

# 2. Custom AI tool with native dependency injection
def verify_token(authorization: str = Header(...)) -> str:
    if not authorization.startswith("Bearer "):
        raise HTTPException(status_code=401, detail="Invalid token")
    return authorization.split(" ")[1]

@mcp.tool(name="order_status", description="Check customer order status")
def check_order(order_id: str, user: str = Depends(verify_token)) -> dict:
    return {"order_id": order_id, "customer": user, "status": "Shipped"}

Shortened. The full README is on GitHub.