MCP Crash Course

What is MCP Crash Course?

A demonstration project showcasing integration of LangChain with Model Context Protocol (MCP) adapters. It implements a system that handles mathematical calculations and weather queries through separate MCP servers. Designed for developers learning MCP and LangChain.

How to use MCP Crash Course?

Clone the repository, create a Python virtual environment, install dependencies from requirements.txt, and add your OpenAI API key to a .env file. Then run python main.py for the single‑server example or python langchain_client.py for the multi‑server example.

Key features of MCP Crash Course

• Integrates LangChain with MCP adapters
• Multiple MCP servers (math and weather)
• Async operation support
• Environment variable configuration
• Both single‑server and multi‑server usage examples

Use cases of MCP Crash Course

• Learning how to integrate MCP with LangChain
• Demonstrating math and weather queries via separate MCP servers
• Prototyping multi‑server agent systems
• Teaching MCP stdio client and MultiServerMCPClient usage
• Reference for building custom MCP adapter examples

FAQ from MCP Crash Course

What are the runtime requirements?

Python 3.x and an OpenAI API key are required. Dependencies include langchain-core, langchain-openai, langchain-mcp-adapters, langgraph, python-dotenv, and mcp.

How do the two main scripts differ?

main.py uses a stdio client to connect to a single math server. langchain_client.py uses MultiServerMCPClient to manage both a math server and a weather server, allowing the agent to choose the appropriate tool.

Does MCP Crash Course require external services?

Yes. The LangChain agent uses OpenAI’s API, so a valid OPENAI_API_KEY must be set in the .env file.

What transport does the project use?

The single‑server example uses stdio client communication. The multi‑server example uses MCP’s MultiServerMCPClient, though the exact transport is not explicitly stated.