armstrong99/Model-Context-Protocol---Fast-API

MCP Demo with FastAPI

This is a toy project to demonstrate how Model Context Protocol (MCP) can work between:

1. A Client Server (simulating an LLM/chatbot).
2. A MCP Action Server (defining actions like "setup a meeting").

The goal is to mimic how an LLM might parse user intent and call out to an MCP server to perform specific actions.

🛠️ Requirements

• Python 3.9+
• Install dependencies:

pip install -r requirements.txt

🚀 Running the Servers

1. Start the MCP Action Server

This server exposes actions (e.g., setup_meeting).

 python3 mcp_server/main.py

2. Start the Client Server

This simulates a chatbot that listens for prompts and calls the MCP server.

cd inference

python3 -m uvicorn main:app --host 0.0.0.0 --port 8000

💬 Example Usage

1. Send a prompt to the client server:

curl -X POST "http://localhost:8000/prompt" \
    -H "Content-Type: application/json" \
    -d '{"prompt": "Please setup a meeting at 12PM"}'

2. The client server will:

   • Parse the text.
   • Detect intent ("setup meeting").
   • Forward the request to the MCP Action Server.
   • Return the MCP server’s response.

📂 Project Structure

.
├── mcp_server/main.py   # MCP Action Server (provides supported actions)
├── inference/main.py   # Client Server (simulates LLM / chatbot)
├── requirements.txt   # Dependencies

⚡ Notes

• This is a toy demo — no real LLM is used.
• Instead of natural language understanding, it uses keyword matching to detect actions.
• You can extend action_server.py to support more actions (e.g., do_xyz, send_sms).