FastMCP Examples

About

This project provides practical demonstrations for building and consuming Model Context Protocol (MCP) servers using Python's FastMCP library. It showcases two primary transport mechanisms: HTTP with a Weather MCP Server integrated with LangChain for dynamic tool discovery by LLMs, and STDIO with a Calculator MCP Server designed for MCP-native hosts. The repository also includes a LangChain client that connects to the HTTP server, illustrating how to convert MCP tools into LangChain's StructuredTool objects for use with OpenAI-compatible models.

Key Features

• HTTP-based MCP Server implementation (Weather API integration)
• STDIO-based MCP Server implementation (Calculator functions)
• LangChain client for dynamic LLM tool discovery and usage
• Dynamic fetching and conversion of MCP tool schemas to LangChain StructuredTools
• Uses FastMCP library for simplified MCP server development
• 0 GitHub stars

Use Cases

• Integrating custom tools and functionalities with Large Language Models (LLMs) via LangChain
• Creating MCP servers accessible over HTTP for web-based clients and services
• Developing MCP-native tools for integration with STDIO-based hosts like Claude Desktop or IDEs