ScriptsMay 11, 2026·2 min read

MCP Python SDK — Build MCP Servers & Clients

Build MCP servers and clients in Python with FastMCP, plus a CLI for dev and install flows; supports stdio, SSE, and Streamable HTTP transports.

MCP Hub
MCP Hub · Community
Agent ready

Safe staging for this asset

This asset is staged first. The copied prompt tells the agent to inspect the staged files and ask before activating scripts, MCP config, or global config.

Stage only · 17/100Policy: stage
Agent surface
Any MCP/CLI agent
Kind
CLI Tool
Install
Stage only
Trust
Trust: Established
Entrypoint
Asset
Safe staging command
npx -y tokrepo@latest install 06ee3ea4-e573-4f6c-8cc4-cef1c1cc8de0 --target codex

Stages files first; activation requires review of the staged README and plan.

Intro

Build MCP servers and clients in Python with FastMCP, plus a CLI for dev and install flows; supports stdio, SSE, and Streamable HTTP transports.

  • Best for: Python teams building their own MCP tools (servers) or integrating MCP into internal agent runtimes
  • Works with: Python, uv/pip workflows, MCP transports (stdio/SSE/streamable HTTP) and FastMCP decorators
  • Setup time: 15 minutes

Quantitative Notes

  • PyPI install group: mcp[cli] (repo docs)
  • GitHub stars + forks (verified): see Source & Thanks
  • Setup time ~15 minutes (Python env + one demo server + inspector)

Practical Notes

A minimal, reliable path is: build one FastMCP server with 1–2 tools, run it in dev mode, then validate with the official inspector. Only after that, add auth/env handling and move to HTTP transports. If you’re using Claude Code, prefer explicit install steps and keep tool inputs narrow and typed.

Safety note: Don’t expose unconstrained exec/file tools over HTTP without an allowlist and auditing; MCP transports are powerful by design.

FAQ

Q: Is this only for servers? A: No. The SDK supports both building servers and writing MCP clients that connect to servers.

Q: Which transport should I pick? A: Use stdio for local tools, and Streamable HTTP for production-like deployments; SSE is also supported.

Q: Do I need uv? A: No. uv is recommended in docs, but pip works too (pip install "mcp[cli]").

🙏

Source & Thanks

GitHub: https://github.com/modelcontextprotocol/python-sdk Owner avatar: https://avatars.githubusercontent.com/u/182288589?v=4 License (SPDX): MIT GitHub stars (verified via api.github.com/repos/modelcontextprotocol/python-sdk): 22,953 GitHub forks (verified via api.github.com/repos/modelcontextprotocol/python-sdk): 3,411

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets

mcp-use — Fullstack SDK for MCP Servers & Apps

mcp-use is a fullstack SDK to build MCP servers and MCP apps across Claude/ChatGPT, with inspector + quickstarts. Verified 9952★; pushed 2026-05-14.

MCP Configs
MCP Hub

mcp-use — Fullstack MCP Framework for AI Agents & Apps

Build MCP servers and apps for ChatGPT, Claude, and any LLM with TypeScript or Python SDK. Includes inspector, cloud deploy, and interactive widgets.

MCP ConfigsScripts
TokRepo精选

Klavis — MCP Integration Platform (Strata + SDK)

Klavis is an MCP integration platform: run hosted or self-hosted Strata servers, with Python/TS SDK + REST for creating tool instances (5.7k★).

MCP Configs
MCP Hub

OpenAI Agents SDK — Build Multi-Agent Systems in Python

Official OpenAI Python SDK for building multi-agent systems with handoffs, guardrails, and tracing. Agents delegate to specialists, enforce safety rules, and produce observable traces. 8,000+ stars.

SkillsCLI Tools
OpenAI
◈Home🔍Search👤Me