🧠 MCP Server with LangChain and AI Tools

This project demonstrates how to build a multi-tool AI assistant using the Model Context Protocol (MCP), LangChain, and Groq’s Qwen model. It includes:

• 📐 A local Math MCP Server
• 🌤️ A simulated Weather MCP Server
• 🤖 A conversational AI agent (MCP client) that talks to both

🧰 Features

• Uses LangChain MCP Adapters to connect tools
• Powered by Groq's Qwen LLM
• Handles local and remote tool servers via MCP
• Interactive CLI chat with tool usage detection

🏁 Prerequisites

• Python >= 3.11
• uv for project/environment management (https://github.com/astral-sh/uv)
• Internet connection for loading LLM (Groq)

⚙️ Setup Instructions

1. Create Project

mkdir mcp_project
cd mcp_project
uv init

Set Python version in .python-version and pyproject.toml to >=3.11

2. Create Virtual Environment

uv venv
source .venv/Scripts/activate

3. Add Dependencies

Create a requirements.txt file:
langchain-mcp-adapters
langchain-groq
langgraph
mcp

Install them

uv add -r requirements.txt

Project Structure

mcp_project/ │ ├── math_server.py # MCP server for math tools ├── weather_server.py # MCP server for weather API simulation ├── client.py # MCP client with AI agent ├── requirements.txt ├── .python-version └── .env # For storing Groq API key (GROQ_API_KEY)

How to Run

1. Run the Weather Server

python weather_server.py

2. Run the Client (Automatically runs math server as sub process)

python client.py

Example Conversation

You: What is the output of 2*3/(4-2)
AI: The result is 3.0

You: What is the weather in New York?
AI: The current weather in New York is sunny.

You: thanks
AI: You're welcome! 😊

Note

The weather server is simulated. Replace it with real API logic if needed.

You can add more MCP servers for documents, search, DBs, etc.

Use .env to store your GROQ_API_KEY.