mcp-client-server-python

MCP Client-Server Python Example

This project demonstrates a simple client-server MCP.

---

What is MCP (Model Context Protocol)?

MCP is an open protocol introduced by Anthropic to enable large language models (LLMs) to interact with external tools, APIs, and resources in a standardized, extensible way.
It facilitates secure, multi-channel communication between AI models and external systems, supporting advanced agentic workflows and tool use.

🔗 Anthropic's announcement
🔗 MCP documentation

---

Features

• MCP Server: Exposes tools (e.g., addition) and resources (e.g., greetings) via SSE.
• MCP Client: Connects to the server, lists available tools, and interacts using OpenAI's chat completions.
• OpenAI Integration: Uses OpenAI's GPT models to process user queries and call server tools as needed.

---

Requirements

• Python 3.12+
• MCP Python SDK
• OpenAI Python SDK
• Uvicorn (for running the server)
• python-dotenv (for loading environment variables)
• uv (fast Python package installer and resolver)

Project Structure

.
├── client.py      # MCP client implementation
├── server.py      # MCP server implementation
├── pyproject.toml # Project metadata and dependencies
├── .env           # Environment variables (not committed)
└── README.md      # This file

Install dependencies with uv:

uv sync

(This will install all dependencies as specified in uv.lock.)

---

Setup

1. Environment Variables

   Create a .env file in the project directory:

   OPENAI_API_KEY=your-openai-api-key
   MCP_SSE_URL=http://localhost:8080/sse
   

2. Start the Server

   uv run server.py --host 0.0.0.0 --port 8080
   

   The server exposes tools and resources via SSE at /sse.

3. Run the Client

   In another terminal:

   uv run client.py
   

   The client will connect to the server, list available tools, and start an interactive chat loop.

---

Usage

• Type your queries in the client prompt.
• The client will use OpenAI to process your query and call server tools if needed.
• Type quit to exit the client.