MCP to LangChain/LangGraph Adapter

What is MCP to LangChain/LangGraph Adapter?

It bridges MCP (Multi‑modal Conversational Procedure) servers and LangChain/LangGraph frameworks, converting MCP tools into LangChain‑compatible tools for use in agents, chains, and graph‑based AI applications.

How to use MCP to LangChain/LangGraph Adapter?

Install the required Python packages (mcp, langchain, langchain‑openai, langgraph, python‑dotenv). Create an MCP server script (e.g., simple_server.py), then instantiate MCPAdapter with the script path. Call get_tools() to retrieve LangChain tools and integrate them into LangChain agents or LangGraph react agents.

Key features of MCP to LangChain/LangGraph Adapter

• Connects to any MCP server via stdio transport.
• Automatically discovers and converts MCP tools to LangChain BaseTool.
• Supports both LangChain agents and LangGraph react agents.
• Manages conversation history with LangGraph memory checkpoints.
• Allows custom environment variables per server process.

Use cases of MCP to LangChain/LangGraph Adapter

• Adding MCP‑exposed tools (e.g., math, weather) to an LLM agent.
• Building conversational agents in LangGraph with persistent state.
• Wrapping legacy or custom tools as MCP servers for LangChain pipelines.

FAQ from MCP to LangChain/LangGraph Adapter

What are the required dependencies?

You need mcp, langchain, langchain‑openai, langgraph, and python‑dotenv installed via pip or pipenv.

How do I connect to an MCP server?

Provide the path to the MCP server script to MCPAdapter; the adapter launches and manages the server process automatically over stdio.

Do I need an API key?

The adapter itself requires no API key. To use OpenAI models in examples, set an OPENAI_API_KEY in a .env file or environment.

What transport does the adapter use?

The provided examples use stdio transport. Other transports are not documented in the README.

How can I debug connection issues?

Add logging to your MCP server script with logging.basicConfig to capture debug output to a file and stream.