FastMCP

FastMCP is a high-level, Pythonic framework for building Model Context Protocol servers (and clients). Instead of hand-wiring JSON-RPC and transports, you decorate plain Python functions — @mcp.tool, @mcp.resource, @mcp.prompt — and FastMCP turns them into a compliant server, generating the schemas the client advertises to the model from your type hints and docstrings.

It is aimed at Python developers who want to expose a capability as an MCP server with minimal ceremony, then grow it toward production. FastMCP's original version (1.0) was incorporated into the official MCP Python SDK; the actively developed standalone framework — now FastMCP 3.x — adds the surrounding machinery: authentication, deployment, server composition, proxying, and generating servers from existing OpenAPI/FastAPI apps.

Highlights

• Decorator-based — @mcp.tool, @mcp.resource, and @mcp.prompt on ordinary functions; schemas are derived from type hints and docstrings.
• Transports built in — run over stdio for local use or Streamable HTTP for remote, without rewriting your handlers.
• Auth & deployment — built-in authentication patterns and deployment helpers for taking a server remote and production-ready.
• Composition & proxying — mount sub-servers and proxy other MCP servers to assemble larger surfaces from smaller ones.
• Generate from existing APIs — produce an MCP server from an OpenAPI spec or a FastAPI app, so an existing service becomes model-accessible.

In an AI-assisted workflow

Define a tool as a typed function and run the server:

from fastmcp import FastMCP
 
mcp = FastMCP("weather")
 
@mcp.tool
def get_weather(city: str, units: str = "c") -> str:
    """Get the current weather for a city. Returns temperature and conditions."""
    data = fetch_weather(city, units)
    return f"{data.temp}° — {data.conditions}"
 
if __name__ == "__main__":
    mcp.run()  # stdio by default; switch to Streamable HTTP for remote

Good to know

FastMCP is free and open source under Apache-2.0. It's the fast path to an MCP server in Python; for the conceptual model see Building an MCP Server, and for taking the result remote and scalable, Deploying a Remote MCP Server.