intro_to_MCP

# Model Context Protocol (MCP) with OpenAI A simple implementation demonstrating how to use Anthropic's Model Context Protocol (MCP) with the OpenAI API. ## What is MCP? Model Context Protocol (MCP) is an open standard by Anthropic that allows large language models to access external data and tools through a standardized interface. It enables models to interact with tools and resources while maintaining context. ## Prerequisites - Python 3.10 or higher - OpenAI API key ## Quick Start Guide ### 1. Install Required Packages ```bash pip install mcp openai ``` ### 2. Set Up Your Project Clone this repository or download the files: - `server.py` - MCP server with example tools - `simple_client.py` - Client that demonstrates OpenAI integration ### 3. Configure Your API Key Update `simple_client.py` with your OpenAI API key: ```python OPENAI_API_KEY = "your_openai_api_key_here" ``` ### 4. Run the Demo ```bash python simple_client.py ``` This will: 1. Start the MCP server with example tools 2. Demonstrate calling the echo tool 3. Demonstrate calling the add tool 4. Show how to chain multiple tools together ## How It Works ### MCP Server (`server.py`) The server defines resources and tools: ```python # Define a tool that echoes a message @mcp.tool() def echo(message: str) -> str: """Echoes back the message provided""" return f"Echo: {message}" # Define a tool that performs addition @mcp.tool() def add(a: int, b: int) -> int: """Adds two numbers together""" return a + b ``` ### OpenAI Integration The client demonstrates how to: 1. Define tool specifications for OpenAI 2. Send function calls to the MCP server 3. Return tool results to OpenAI Example tool definition for OpenAI: ```python { "type": "function", "function": { "name": "echo", "description": "Echoes back the message provided", "parameters": { "type": "object", "properties": { "message": {"type": "string"} }, "required": ["message"] } } } ``` ## Creating Your Own Tools To add new tools to the MCP server: 1. Add a new function to `server.py` with the `@mcp.tool()` decorator 2. Define the input parameters and return type 3. Add the corresponding tool definition in your OpenAI client Example: ```python @mcp.tool() def multiply(a: int, b: int) -> int: """Multiplies two numbers together""" return a * b ``` ## Resources - [MCP Documentation](https://modelcontextprotocol.io/) - [OpenAI Function Calling API](https://platform.openai.com/docs/guides/function-calling)