FastMCP: Autonomous LangChain Documentation Agent

An advanced implementation of the Model Context Protocol (MCP) for intelligent documentation exploration. Powered by LangChain, LangGraph, and Groq LLMs.

---

Project Overview

This project implements a state-of-the-art LangChain Documentation Agent using the high-level FastMCP framework. It bridges the gap between static documentation and interactive AI by allowing an autonomous agent to search, fetch, and summarize technical docs in real-time through the Model Context Protocol.

Strategic Architecture

• FastMCP Server: A specialized MCP server (server/scraper.py) that exposes tools for semantic search and content extraction from the LangChain documentation ecosystem.
• LangGraph Orchestrator: An agentic loop implemented in agent/agent.py that utilizes LangGraph's native handling of JSON-schema tools to solve standard langchain_classic compatibility issues.
• Groq Inference: Utilizes Llama 3.3 (70B) via Groq for ultra-fast, high-reasoning capabilities, enabling the agent to execute complex multi-step research tasks.

---

🛠️ Key Features

• Semantic Documentation Search: Deep-search capabilities across LangChain's official guides, tutorials, and API references.
• Autonomous Tool Selection: The agent intelligently decides when to list topics, search for specific keywords, or fetch full page content.
• Conflict Resolution: Implements the definitive fix for MCP tool-input schema mismatches by leveraging LangGraph's object-passing architecture.
• Background Orchestration: Automated background server startup with health polling ensures the agent only starts once the MCP bridge is ready.

---

📂 Project Structure

fastmcp/
├── agent/            # LangGraph Agent logic & MCP Client bridge
├── server/           # FastMCP Server implementation & Scraper logic
├── run.py            # Unified entry point for Server + Agent orchestration
├── requirements.txt  # Project dependencies
└── .env              # API keys and server configuration

---

⚙️ Setup & Execution

1. Configure Environment

Create a .env file with your Groq API Key:

GROQ_API_KEY=gsk_your_key_here
MCP_SERVER_HOST=localhost
MCP_SERVER_PORT=8000

2. Install Dependencies

python -m venv env
source env/Scripts/activate  # Windows
pip install -r requirements.txt

3. Launch the System

python run.py

This command will:

1. Start the FastMCP Server in a background daemon thread.
2. Wait for the server to pass its Health Check.
3. Initialize the LangGraph Agent.
4. Begin an interactive session or execute pre-defined documentation queries.

---

"Leveraging the Model Context Protocol to build agents that truly understand their tools."