Turn LangChain Docs into a Live Dev Assistant (LangGraph Dev Navigator MCP Demo)

Description

Most AI helpers hallucinate about LangChain and LangGraph. In this demo, I show how **LangGraph Dev Navigator** uses an MCP server + RAG to answer: “How do I reduce latency for agent / multi-agent workflows in the *official* LangChain / LangGraph / LangSmith docs?” What you’ll see: - Real developer question about latency (parallel tool calls, batching, etc.) - MCP server calling tools (RAG + code search) over official docs only - Pulling concrete knobs like `max_concurrency`, `aevaluate`, `LANGSMITH_TEST_CACHE`, and LangGraph `Command` handoffs - A final answer that is citation-backed and honest about missing docs (no fake “parallel tool calls” guide) 🌱 What is LangGraph Dev Navigator? An MCP server that turns LangChain / LangGraph documentation into an executable, citation-backed assistant for real developer questions and coding tasks. We keep up with upstream changes and answer against the versions you actually have installed. 🚀 Join the early-access waitlist: https://langgraph-dev-navigator.replit.app #langchain #langgraph #MCP #agents #RAG #AItools #developer github: https://github.com/botingw linkedin: https://www.linkedin.com/in/bo-ting-wang/ X: https://x.com/BoTingWang51306