langchain-mcp-adapters

What it is

This library provides a lightweight wrapper that makes Anthropic Model Context Protocol (MCP) tools compatible with LangChain and LangGraph. It includes a client implementation to connect to multiple MCP servers and load tools from them.

How it works

• Converts MCP tools into LangChain tools usable with LangGraph agents.
• Includes a client to connect to MCP servers and load tools from them.
• Supports multiple server connections via a MultiServerMCPClient.
• Added support for streamable HTTP transport and runtime headers on supported transports.

Getting started

pip install langchain-mcp-adapters

pip install langchain-mcp-adapters langgraph "langchain[openai]"

export OPENAI_API_KEY=<your_api_key>

Server example

# math_server.py
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("Math")

@mcp.tool()
def add(a: int, b: int) -> int:
    """Add two numbers"""
    return a + b

@mcp.tool()
def multiply(a: int, b: int) -> int:
    """Multiply two numbers"""
    return a * b

if __name__ == "__main__":
    mcp.run(transport="stdio")

Client example (single server)

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

server_params = StdioServerParameters(
    command="python",
    args=["/path/to/math_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)
        agent = create_agent("openai:gpt-4.1", tools)
        agent_response = await agent.ainvoke({"messages": "what's (3 + 5) x 12?"})

Multiple MCP servers

from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent

client = MultiServerMCPClient(
    {
        "math": {
            "command": "python",
            "args": ["/path/to/math_server.py"],
            "transport": "stdio",
        },
        "weather": {
            "url": "http://localhost:8000/mcp",
            "transport": "http",
        }
    }
)
tools = await client.get_tools()
agent = create_agent("openai:gpt-4.1", tools)
math_response = await agent.ainvoke({"messages": "what's (3 + 5) x 12?"})
weather_response = await agent.ainvoke({"messages": "what is the weather in nyc?"})

Streamable HTTP

cd examples/servers/streamable-http-stateless/
uv run mcp-simple-streamablehttp-stateless --port 3000

from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
from langchain.agents import create_agent
from langchain_mcp_adapters.tools import load_mcp_tools

async with streamablehttp_client("http://localhost:3000/mcp") as (read, write, _):
    async with ClientSession(read, write) as session:
        await session.initialize()
        tools = await load_mcp_tools(session)
        agent = create_agent("openai:gpt-4.1", tools)
        math_response = await agent.ainvoke({"messages": "what's (3 + 5) x 12?"})

Passing runtime headers

from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent

client = MultiServerMCPClient(
    {
        "weather": {
            "transport": "http",
            "url": "http://localhost:8000/mcp",
            "headers": {
                "Authorization": "Bearer YOUR_TOKEN",
                "X-Custom-Header": "custom-value"
            },
        }
    }
)
tools = await client.get_tools()
agent = create_agent("openai:gpt-4.1", tools)
response = await agent.ainvoke({"messages": "what is the weather in nyc?"})

Tool error handling

MCP distinguishes tool execution errors from transport failures. By default, handle_tool_errors is true. You can disable legacy behavior by passing handle_tool_errors=False when constructing the client or loading tools.

client = MultiServerMCPClient({...})
tools = await client.get_tools()  # handle_tool_errors=True by default

The error content blocks are preserved verbatim on the ToolMessage; if no content is provided by MCP, a minimal placeholder text block is used.

Using with LangGraph StateGraph

from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.graph import StateGraph, MessagesState, START
from langgraph.prebuilt import ToolNode, tools_condition

from langchain.chat_models import init_chat_model
model = init_chat_model("openai:gpt-4.1")

client = MultiServerMCPClient(
    {
        "math": {
            "command": "python",
            "args": ["./examples/math_server.py"],
            "transport": "stdio",
        },
        "weather": {
            "url": "http://localhost:8000/mcp",
            "transport": "http",
        }
    }
)
tools = await client.get_tools()

def call_model(state: MessagesState):
    response = model.bind_tools(tools).invoke(state["messages"])
    return {"messages": response}

builder = StateGraph(MessagesState)
builder.add_node(call_model)
builder.add_node(ToolNode(tools))
builder.add_edge(START, "call_model")
builder.add_conditional_edges(
    "call_model",
    tools_condition,
)
builder.add_edge("tools", "call_model")
graph = builder.compile()
math_response = await graph.ainvoke({"messages": "what's (3 + 5) x 12?"})
weather_response = await graph.ainvoke({"messages": "what is the weather in nyc?"})