LangChain MCP – Restaurant Example

A demonstration of using Model Context Protocol (MCP) to expose a restaurant interaction tool that integrates with LangChain.

---

📁 Repo Structure

• restaurant_mcp.py – MCP server exposing restaurant-related tools.
• restaurant_data.py – Mock dataset (restaurant listings and menus).
• mcp_client.py – Python client to connect and call MCP server tools.
• requirements.txt – Python dependencies for server & client.
• .env – Environment variables files, will have you OPENAI_API_KEY

---

🚀 Setup

1. Clone and install dependencies

   git clone https://github.com/pk-pkulkarni/langchain_mcp.git
   cd langchain_mcp
   pip install -r requirements.txt
   

2. Run the MCP server Open a terminal and start the server:

   python restaurant_mcp.py

⚠️ Make sure restaurant_mcp.py is running as a server in your terminal before proceeding.

3. Run the client(For testing)
   In a separate terminal, run:

python mcp_client.py

4. Run FastAPI Server
   Run File with uvicorn:

uvicorn mcp_client_api_chatbot:app --reload --port 8000

5. Open chatbot.html
   Examples:

Give me veg menu
Give me all soups