fastmcp-server

FastMCP 3.0 Server Development

Complete reference for building production-ready MCP (Model Context Protocol) servers with FastMCP 3.0 - the fast, Pythonic framework for connecting LLMs to tools and data.

When to use this skill

Use FastMCP Server when:

• Creating a new MCP server in Python
• Adding tools, resources, or prompts to an MCP server
• Implementing authentication (OAuth, OIDC, token verification)
• Setting up middleware for logging, rate limiting, or authorization
• Configuring providers (local, filesystem, skills, custom)
• Building production MCP servers with telemetry and storage
• Upgrading from FastMCP 2.x to 3.0

Key areas covered:

• Tools & Resources (CORE): Decorators, validation, return types, templates
• Context & DI (CORE): MCP context, dependency injection, background tasks
• Authentication (SECURITY): OAuth, OIDC, token verification, proxy patterns
• Authorization (SECURITY): Scope-based and role-based access control
• Middleware (ADVANCED): Request/response pipeline, built-in middleware
• Providers (ADVANCED): Local, filesystem, skills, and custom providers
• Features (ADVANCED): Pagination, sampling, storage, OpenTelemetry, versioning

Quick reference

Core patterns

Create a server with tools:

from fastmcp import FastMCP

mcp = FastMCP("MyServer")

@mcp.tool
def add(a: int, b: int) -> int:
    """Add two numbers"""
    return a + b

Create a resource:

@mcp.resource("data://config")
def get_config() -> dict:
    """Return server configuration"""
    return {"version": "1.0", "debug": False}

Create a resource template:

@mcp.resource("users://{user_id}/profile")
def get_user_profile(user_id: str) -> dict:
    """Get a user's profile by ID"""
    return fetch_user(user_id)

Create a prompt:

@mcp.prompt
def review_code(code: str, language: str = "python") -> str:
    """Review code for best practices"""
    return f"Review this {language} code:\n\n{code}"

Run the server:

if __name__ == "__main__":
    mcp.run()

# Or with transport options:
# mcp.run(transport="sse", host="0.0.0.0", port=8000)

Using context in tools