Claude Desktop Cursor Windsurf (Cascade)Cline / Roo Code OpenAI Codex CLI Antigravity Zed Editor LangChain / LangGraph

How to Setup MCP in LangChain / LangGraph

SDK / Framework

Step-by-step setup guide for connecting Model Context Protocol (MCP) servers to LangChain / LangGraph by LangChain.

Configuration File: Programmatic

Code Integration

Example Programmatic Template

Open in Config Generator

from langchain_mcp_adapters.tools import load_mcp_tools
async with stdio_client(server_params) as (read, write):
    tools = await load_mcp_tools(read, write)

Installation Steps

Locate Programmatic

Find the configuration file at Code Integration.

Add MCP server entries

Open the file and add your server definitions inside the configuration schema.

Set environment variables

Provide any required API keys, database URLs, or access tokens in the environment section.

Save and reload

Save Programmatic and hot-reload or restart LangChain / LangGraph.

Verify in chat

Ask LangChain / LangGraph to list tools or execute a task with the connected MCP server.

LangChain / LangGraph Protocol Support

Stdio Transport Supported

stdio_client transport adapter

SSE / HTTP Transport Supported

sse_client transport adapter

Tool Calling Supported

StructuredTool conversion

Resources API Supported

Document loader wrappers

Prompts API Supported

ChatPromptTemplate conversion

Roots Boundary Unsupported

Custom memory management

Dynamic Hot-Reload Supported

Client runtime connection pool

LangChain / LangGraph Setup FAQ

Where is the LangChain / LangGraph MCP config file located?

On Windows: Code Integration. On macOS: Code Integration. On Linux: Code Integration.

Does LangChain / LangGraph support MCP tools and stdio transports?

LangChain / LangGraph supports stdio transports and supports autonomous tool calling.

How do I reload MCP servers in LangChain / LangGraph?

Client runtime connection pool