FastMCP

Overview

FastMCP is to MCP servers what FastAPI is to web APIs. Write a Python function with type hints, decorate it, and FastMCP handles everything — serialization, validation, transport, and the full MCP protocol handshake. With 24k+ stars and backing from Prefect, it’s the standard way to build custom MCP servers.

How It Works

1. Define tools — Write Python functions with type annotations
2. Decorate — Add @mcp.tool() to expose via MCP
3. Run — FastMCP handles protocol and transport
4. Connect — Any MCP client discovers your tools

Use Cases

• Custom internal tools — Expose company APIs to AI agents
• Database wrappers — Natural language database access
• Business logic — Domain-specific agent operations
• API bridges — Convert REST APIs into MCP tools

Getting Started

from fastmcp import FastMCP

mcp = FastMCP("My Tools")

@mcp.tool()
def get_weather(city: str) -> str:
    """Get current weather for a city."""
    return f"Weather in {city}: 72°F, sunny"

mcp.run()

Example

@mcp.tool()
def check_stock(product_id: str) -> dict:
    """Check inventory levels for a product."""
    stock = db.query(f"SELECT * FROM inventory WHERE id = %s", (product_id,))
    return {"product": product_id, "quantity": stock.quantity}

Alternatives

• FastAPI MCP — Expose existing FastAPI endpoints as MCP tools
• MCP SDK — Low-level official MCP SDK from Anthropic
• MCP Use — Full-stack MCP application framework