Langchain Mcp

io.github.baixiangerPublished by GitHub account baixiangerv2.0.0Updated Oct 24, 2026

LangChain/LangGraph/DeepAgents knowledge + LangGraph agent debugging (Polly-like)

VerifiedSTDIODesktop onlyKnowledge & MemoryDeveloper Tools

Overview

AI-generated overview

A local MCP server that lets an assistant search LangChain, LangGraph, LangSmith and DeepAgents docs and code, and inspect local LangGraph agent runs.

What you can do
Search LangChain, LangGraph, LangSmith and DeepAgents documentation and tutorials, and search the LangChain, LangGraph and DeepAgents source code (requires login). Debug local LangGraph agents by listing threads, runs and traces, and reading thread details, state and checkpoints. Ask for analysis of why a run failed and get suggestions for improving an agent's prompts.
When to use it
Useful when building with LangChain, LangGraph or DeepAgents and you want documentation and source lookups inside the assistant, or when debugging a LangGraph agent running on a local langgraph dev server.
Requirements
Runs as a local npm package installed globally and registered with the assistant over stdio. Search tools require running the login command first; debugging tools need a local langgraph dev server, connected at localhost:2024 by default. Desktop only.
Before you install
Search tools require an account login and consume paid credits (about $5 of free credits included); debugging tools are listed as free. The server reads local LangGraph agent state, runs and traces, which may contain sensitive prompt or data content.

Installation

In SourceWeft

  1. Open Langchain Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

langchain-mcp

Give Claude superpowers for LangChain development

[npm version] [License: MIT]

Website · GitHub · Sponsor

Features

Knowledge Search

Search LangChain, LangGraph, LangSmith & DeepAgents documentation and source code instantly from Claude.

LangGraph Agent Debugging (New in v2!)

Debug your LangGraph agents like Polly - connect to your local langgraph dev server and:

  • List threads, runs, and traces
  • Inspect agent state and checkpoints
  • Ask "Why did this fail?" and get AI-powered analysis
  • Get suggestions to improve your agent's prompts

Install

bash
npm i -g langchain-mcplangchain-mcp loginclaude mcp add langchain-mcp -- langchain-mcp
Claude Desktop config

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

json
{  "mcpServers": {    "langchain-mcp": {      "command": "langchain-mcp"    }  }}

Tools

Search (requires login)

ToolDescription
search_docsSearch documentation and tutorials
search_langchain_codeSearch LangChain source code
search_langgraph_codeSearch LangGraph source code
search_deepagents_codeSearch DeepAgents source code

Debug LangGraph Agents (free, no login)

ToolDescription
langgraph_list_threadsList threads from local server
langgraph_get_threadGet thread details
langgraph_get_thread_stateGet state/checkpoint
langgraph_list_runsList runs for a thread
langgraph_get_runGet run trace details

Debug Your Agents

bash
# Start your LangGraph serverlanggraph dev

Then ask Claude:

  • "List my LangGraph threads"
  • "What went wrong with thread abc123?"
  • "Summarize the last run"
  • "How can I improve this agent?"

Connects to localhost:2024 by default.

Pricing

FeatureCost
Doc/Code Search$5 free credits (~2000 searches)
LangGraph DebuggingFree forever

License

MIT

Source: packages/mcp-server/README.md at commit 78521ea

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v2.0.0LatestSep 16, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.