日本語版はこちらをご覧ください。

MCP Azure Client

A Python-based Model Context Protocol (MCP) client specifically designed to interact with Azure. This application provides a convenient way to connect to Azure MCP servers and call available tools through an interactive command-line interface.

Features

• Connects to Azure MCP servers using the MCP protocol
• Lists available tools with their descriptions and required parameters
• Allows interactive execution of tools with parameter input
• Packaged in Docker for consistent deployment across environments

Requirements

• Python 3.12+
• Required Python packages:
  • mcp: Model Context Protocol library
  • uv: Dependency management
• Docker (for containerized execution)
• Node.js & npm (for Azure MCP server connection)

Usage

Docker Compose (Recommended)

The application is containerized for ease of use and is best run using Docker Compose, which manages all dependencies and environment setup automatically.

1. From the project directory, start the service:

   docker compose up -d
   

   After the service starts for the first time, enter the container and log in with Azure CLI:

   docker compose exec mcp-client az login
   

2. This will build the container, start the application, and show the logs directly in your terminal.

   docker compose exec mcp-client python3 mcp-client.py
   

3. When the application starts, it will automatically connect to the Azure MCP server and show the client interface.

4. To stop the service, press Ctrl+C and then run:

   docker compose down
   

Local Installation

If you prefer to run the application locally:

1. Install dependencies:

   pip install -r requirements.txt
   

2. Make sure npm is installed on your system

3. Run the client:

   python mcp-client.py
   

How It Works

1. The application initializes and connects to an Azure MCP server using npx to run the @azure/mcp package
2. Once connected, you can:
   • List all available tools (option 1)
   • Call a specific tool by number (option 2)
   • Exit the application (option 3)
3. When calling a tool, you'll be prompted to input required and optional parameters

Code Overview

The mcp-client.py file contains:

• MCPClient class: Handles server connection logic
• list_and_call_tools function: Manages the interactive CLI interface
• Helper functions for unpacking tool definitions and displaying them
• Main function that orchestrates the application flow

Project Structure

.
├── Dockerfile            # Container definition
├── docker-compose.yml    # Docker Compose configuration
├── mcp-client.py         # Main application code
└── requirements.txt      # Python dependencies