MCP (Model Context Protocol) Integration Examples

This repository demonstrates the usage of Model Context Protocol (MCP) with various integrations including a custom calculator server, GitHub, and Google Maps.

What is MCP?

Model Context Protocol (MCP) is a protocol that enables AI models to interact with external tools and services. It provides a standardized way for AI models to:

• Execute tools and functions
• Access resources
• Generate prompts
• Interact with external services

Project Structure

├── server.py # MCP server (custom creation)
├── client_server.py # MCP Client
├── 1)maps.py # Google Maps MCP integration
├── 2)github.py # GitHub MCP integration
└── requirements.txt # Project dependencies

Features

1. Custom Calculator Server

• Implements basic arithmetic operations
• Demonstrates MCP tool creation
• Shows resource and prompt handling

2. GitHub Integration

• List repository commits
• Uses GitHub Personal Access Token for authentication
• Demonstrates environment variable handling

3. Google Maps Integration

• Search for places using Google Maps API
• Configurable search radius
• Environment variable based API key management

Setup

1. Install dependencies:

pip install -r requirements.txt

MCP Tools Types

1. Tools: Functions that perform specific actions

   • Defined using @mcp.tool decorator
   • Can accept parameters and return values

2. Resources: Static or dynamic data sources

   • Defined using @mcp.resource decorator
   • Accessed using resource URLs

3. Prompts: Template-based text generation

   • Defined using @mcp.prompt decorator
   • Can include dynamic content

Best Practices

1. Always use environment variables for sensitive data
2. Implement proper error handling
3. Use type hints for better code clarity
4. Document your tools with clear docstrings
5. Keep API keys secure and never commit them to version control

Usage Examples

Calculator

Server side ( initialize )

@mcp.tool(name="add")
def add(a: int, b: int): return a + b

Running server.py file, it will up and run the MCP server.

Client side ( utilize )

result = await session.call_tool("add", arguments={"a": 5, "b": 3})

Running client_server.py file, it will connect to the MCP server.