MCP Multi-Server Client Project

March 25, 2025 · View on GitHub

A multi-server client project based on MCP (Model Control Protocol), supporting simultaneous connections to multiple servers and OpenAI API integration.

Project Structure

mcp_project/
├── config/                   # Configuration directory
│   ├── servers.json          # Server configuration file (JSON format)
│   └── api_config.json       # API configuration file (JSON format)
├── src/                      # Source code directory
│   ├── __init__.py           # Package initialization
│   └── mcp_project/          # Main project package
│       ├── __init__.py       # Package initialization
│       ├── core/             # Core modules
│       │   ├── __init__.py   
│       │   ├── multi_server_client.py  # Multi-server client class
│       │   └── server_connection.py    # Server connection class
│       ├── servers/          # Server script directory
│       │   ├── __init__.py
│       │   ├── calculator.py           # Calculator server example
│       │   ├── fileprocessor.py        # File processing server
│       │   ├── python_excutor.py       # Python code execution server
│       │   └── shell_processor.py      # Shell command processor
│       └── utils/            # Utility functions
│           ├── __init__.py
│           └── load_config.py          # Configuration loading utilities
├── run.py                    # Startup script
├── pyproject.toml            # Project configuration
├── requirements.txt          # Dependencies
├── .env.example              # Example environment variables (template)
├── .env                      # Environment variables file (optional)
└── README.md                 # Project documentation

Features

  1. Multi-Server Support: Connect to multiple MCP servers simultaneously based on JSON configuration
  2. Intelligent Tool Routing: Automatically find servers that provide specific tools
  3. OpenAI API Integration: Use OpenAI-compatible API interfaces (e.g., Qwen models)
  4. Interactive Interface: Provides an interactive command-line interface
  5. Configurable API Settings: Set API keys, models, and parameters via configuration files or environment variables
  6. Environment Variable Support: Load configurations from .env files and system environment variables
  7. Code Execution: Python code execution with captured output
  8. File Processing: Read and write files
  9. Shell Command Execution: Execute shell commands

Installation

# Clone the repository
git clone https://github.com/yourusername/mcp_project.git
cd mcp_project

# Create and activate virtual environment (optional)
uv venv --python 3.12
source .venv/bin/activate  # Linux/Mac
# OR
.venv\Scripts\activate     # Windows

# Install dependencies
uv pip install -r requirements.txt

Configuration

Environment Variables

You can set environment variables in two ways:

  1. System Environment Variables: Set them in your operating system
  2. .env File: Create a .env file in the project root directory

Supported environment variables:

OPENAI_API_KEY=your-api-key
OPENAI_BASE_URL=https://api.siliconflow.cn/v1
OPENAI_MODEL_NAME=Qwen/Qwen2.5-7B-Instruct
OPENAI_TEMPERATURE=0.7
OPENAI_TOP_P=1.0
OPENAI_MAX_TOKENS=1000
OPENAI_TOOL_CHOICE=auto

Example .env file:

OPENAI_API_KEY=sk-your-api-key-here
OPENAI_MODEL_NAME=gpt-3.5-turbo
OPENAI_TEMPERATURE=0.5

Server Configuration

Edit the config/servers.json file to configure servers:

{
  "mcpServers": {
    "calculator": {
      "command": "python",
      "args": [
        "src/mcp_project/servers/calculator.py"
      ],
      "name": "Calculator Server",
      "description": "Provides calculation functionality",
      "enable": true
    },
    "python_executor": {
      "command": "python",
      "args": [
        "src/mcp_project/servers/python_excutor.py"
      ],
      "name": "Python Executor",
      "description": "Executes Python code and captures output",
      "enable": true
    },
    "file_processor": {
      "command": "python",
      "args": [
        "src/mcp_project/servers/fileprocessor.py"
      ],
      "name": "File Processor",
      "description": "Provides file reading and writing functionality",
      "enable": true
    },
    "shell_processor": {
      "command": "python",
      "args": [
        "src/mcp_project/servers/shell_processor.py"
      ],
      "name": "Shell Processor",
      "description": "Executes shell commands",
      "enable": true
    }
  }
}

Each server configuration contains the following fields:

  • command: Execution command (e.g., python or node)
  • args: Command arguments array, usually starting with the server script path
  • name: Display name (optional)
  • description: Description (optional)
  • enable: Whether the server is enabled (optional, defaults to true)

API Configuration

Edit the config/api_config.json file to configure the OpenAI API:

{
  "openai_api": {
    "api_key": "your-api-key",
    "base_url": "https://api.siliconflow.cn/v1",
    "model_name": "Qwen/Qwen2.5-7B-Instruct",
    "parameters": {
      "temperature": 0.7,
      "top_p": 1.0,
      "max_tokens": 1000,
      "tool_choice": "auto",
      "timeout": 60
    }
  }
}

Note:

  • If the api_key field is empty, the system will use the OPENAI_API_KEY environment variable
  • If the configuration file doesn't exist, the system will use environment variables or default values
  • Configuration priority: Config file > Environment variables > Default values

Running the Client

# Start with default configuration
python run.py

# Specify custom configuration paths
python run.py --servers path/to/servers.json --api path/to/api_config.json

The client provides an interactive command line interface. Type your queries and the client will process them using the OpenAI API and MCP servers.

Available Servers and Tools

Calculator Server

  • add: Calculate the sum of multiple numbers
  • multiply: Calculate the product of multiple numbers
  • compare: Compare two numbers

Python Executor

  • execute_python_code: Execute Python code and capture stdout output
  • execute_with_globals: Execute Python code with a customizable global namespace

File Processor

  • read_file: Read file content
  • write_file: Write content to a file
  • list_files: List files in a directory

Shell Processor

  • execute_shell_command: Execute shell commands