LangChain 1.4 ships built-in MCP support with MCPAdapter

What you get

Released Sep 1, 2026 in langchain v1.4.0: MCP support now ships inside LangChain as langchain.mcp (built on FastMCP), replacing the separate langchain-mcp-adapters package.

Prerequisites

Python 3.10+ and langchain[mcp]>=1.4.0
An Anthropic API key (for the example model)
Note: langchain.mcp is in beta and emits a LangChainBetaWarning on import

Steps

Install with the mcp extra plus a model integration:

Shell

pip install "langchain[mcp]" langchain-anthropic
export ANTHROPIC_API_KEY="your-api-key"

Point MCPAdapter at any MCP server URL. This example uses LangChain's public docs server (no key needed):

Python

import asyncio
from langchain.agents import create_agent
from langchain.mcp import MCPAdapter

async def main():
    async with MCPAdapter("https://docs.langchain.com/mcp") as adapter:
        tools = await adapter.list_tools()
        agent = create_agent("claude-sonnet-5", tools)
        return await agent.ainvoke(
            {
                "messages": [
                    {
                        "role": "user",
                        "content": "How do I add short-term memory to a LangChain agent?",
                    }
                ]
            }
        )

result = asyncio.run(main())
print(result["messages"][-1].content)

Migrating from langchain-mcp-adapters? MultiServerMCPClient is replaced by MCPAdapter; pass an {"mcpServers": {...}} config dict to connect several servers behind one adapter.

Test it

Run the script. The agent should call the search_docs_by_lang_chain tool and answer with a summary of the short-term memory docs.