langchain-mcp-adapters

What it is and what it does

This library bridges Anthropic's Model Context Protocol (MCP) and LangChain by wrapping MCP tools so they work with LangChain agents and LangGraph workflows. It provides two main entry points: a tool loader that converts a single MCP session's tools into LangChain-compatible objects, and a multi-server client that manages connections to multiple MCP servers and aggregates their tools. The library handles both stdio and HTTP transports, supports custom headers for authentication, and distinguishes between tool execution errors (which it returns as error messages to the agent) and protocol failures.

Typical usage involves creating an MCP server with decorated functions, then on the client side instantiating either a direct session or a MultiServerMCPClient, loading tools from it, and passing those tools to a LangChain agent or LangGraph workflow. The library is actively maintained and requires Python 3.10 or later, with dependencies on langchain-core, mcp, and typing-extensions.

Use it for:

• Build LangGraph agents that call external MCP-based tools (e.g., math operations, weather lookups) as part of agentic workflows
• Aggregate tools from multiple MCP servers and expose them to a single LangChain agent without managing individual connections
• Add authentication headers to MCP server connections for secure tool access in production environments
• Convert existing MCP tool definitions into LangChain tool objects for use with OpenAI or other LLM-based agents

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Converts Anthropic Model Context Protocol (MCP) tools into LangChain-compatible tools for use with LangGraph agents, and provides a client to connect to multiple MCP servers and load their tools.

Yes. The package is actively maintained, has no known vulnerabilities, installs with low friction, and solves a clear integration problem between two widely-used frameworks. Use it if you have MCP tools and want to use them in LangChain or LangGraph agents.

Install

langchain-mcp-adapters on PyPI

pip install langchain-mcp-adapters

uv add langchain-mcp-adapters

poetry add langchain-mcp-adapters

Installing langchain-mcp-adapters

Before you install

Low install friction with a pure Python wheel. Active maintenance: released 8 days ago with 3628 repository stars and ongoing commits. Requires Python 3.10 or later.

License in practice

MIT license permits unrestricted use, modification, and distribution with minimal obligations—suitable for commercial and proprietary projects.

Quickstart

pip install langchain-mcp-adapters

from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client
from langchain_mcp_adapters.tools import load_mcp_tools

server_params = StdioServerParameters(command="python", args=["/path/to/server.py"])
async with stdio_client(server_params) as (read, write):
    async with ClientSession(read, write) as session:
        await session.initialize()
        tools = await load_mcp_tools(session)

Requires Python 3.10 or later; MCP server must be running or accessible at the specified path/URL.

Verify before relying

• Whether the library handles all MCP transport types (stdio, http, streamable-http, sse) equally well or if some have known limitations
• Performance characteristics when connecting to many MCP servers simultaneously
• Compatibility guarantees with specific versions of langchain-core and mcp

Package facts

Evidence: langchain_mcp_adapters-0.3.2-py3-none-any.whl

Tags

More Application Frameworks packages