LangChain Agent with MCP Servers

A LangChain agent using MCP Adapters for tool integration with Model Context Protocol (MCP) servers.

Overview

This project demonstrates how to build a LangChain agent that uses the Model Context Protocol (MCP) to interact with various services:

• Tavily Search: Web search and news search capabilities
• Weather: Mock weather information retrieval
• Math: Mathematical expression evaluation

The agent uses LangGraph's ReAct agent pattern to dynamically select and use these tools based on user queries.

Features

• Graceful Shutdown: All MCP servers implement proper signal handling for clean termination
• Subprocess Management: The main agent tracks and manages all MCP server subprocesses
• Error Handling: Robust error handling throughout the application
• Modular Design: Easy to extend with additional MCP servers

Graceful Shutdown Mechanism

This project implements a comprehensive graceful shutdown system:

1. Signal Handling: Captures SIGINT and SIGTERM signals to initiate graceful shutdown
2. Process Tracking: The main agent maintains a registry of all child processes
3. Cleanup Process: Ensures all subprocesses are properly terminated on exit
4. Shutdown Flags: Each MCP server has a shutdown flag to prevent new operations when shutdown is initiated
5. Async Cooperation: Uses asyncio to allow operations in progress to complete when possible

Installation

# Clone the repository
git clone https://github.com/yourusername/langchain-mcp.git
cd langchain-mcp

# Create a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -e .

Configuration

Create a .env file in the project root with the following variables:

OPENAI_API_KEY=your_openai_api_key
TAVILY_API_KEY=your_tavily_api_key

Usage

Run the agent from the command line:

python src/agent.py

The agent will prompt for your query and then process it using the appropriate tools.

Development

To add a new MCP server:

1. Create a new file in src/mcpserver/
2. Implement the server with proper signal handling
3. Update src/mcpserver/__init__.py to expose the new server
4. Add the server configuration to src/agent.py

License

MIT