MCP Bridge

🚀 MCP-Bridge

MCP-Bridge is a bridge tool that connects the OpenAI API and the inference engine. It adds all available MCP tool definitions to the request, forwards the request to the inference engine, processes the results of tool calls, and finally returns the results to the OpenAI API.

🚀 Quick Start

MCP-Bridge serves as a crucial link between the OpenAI API and the inference engine. It enriches requests with all available MCP tool definitions, routes them to the inference engine, manages tool call outcomes, and relays the results back to the OpenAI API.

✨ Features

• Support for Multiple MCP Servers: Multiple MCP servers can be configured, each with different commands and parameters.
• API Key Authentication: Protect the server by configuring API keys to ensure that only authorized users can access the service.
• Flexible Sampling Settings: Custom sampling timeout and model performance parameters (such as intelligence, cost, and speed) are supported.
• Network Configuration: The server's binding address and port can be adjusted according to requirements.
• Logging: Different levels of log output are supported for easy debugging and monitoring.

📦 Installation

1. Download the Code

Clone the project repository from GitHub:

git clone https://github.com/your-repository.git
cd mcp-bridge

2. Configure the Environment

Edit the config.json file and fill in the inference server address, MCP server commands, and other parameters as needed. For example:

{
    "inference_server": {
        "base_url": "http://localhost:8000/v1",
        "api_key": "None"
    },
    "mcp_servers": {
        "fetch": {
            "command": "uvx",
            "args": ["mcp-server-fetch"]
        }
    }
}

3. Start the Service

Run the following command to start MCP-Bridge:

python main.py

📚 Documentation

1. Basic Configuration

• Inference Server: Specify the API address and key of the inference engine.
• MCP Server: Define available tools and their calling commands.

2. Security Settings

Enable API Key authentication:

{
    "security": {
        "auth": {
            "enabled": true,
            "api_keys": [
                {"key": "your-api-key"}
            ]
        }
    }
}

3. Network Configuration

Adjust the server's binding address and port:

{
    "network": {
        "host": "0.0.0.0",
        "port": 9090
    }
}

4. Logging Settings

Control the log output level:

{
    "logging": {
        "log_level": "DEBUG"
    }
}

💻 Usage Examples

Basic Usage

Send a simple request to MCP-Bridge, including a tool call:

POST /v1/chat/completions HTTP/1.1
Content-Type: application/json
Authorization: Bearer your-api-key

{
    "model": "gpt-4",
    "messages": [
        {
            "role": "user",
            "content": "Please list all available MCP tools."
        }
    ],
    "tools": {
        "enabled": true
    }
}

Advanced Usage

MCP-Bridge returns the results of tool calls, forwards them to the inference engine, and finally returns the response to the OpenAI API.

🔧 Technical Details

mcp-bridge/
├── main.py          # Main program entry point
├── config.json      # Configuration file
├── README.md        # User documentation
└── requirements.txt # Dependency management file

🤝 Contribution and Support

If you encounter any issues or have suggestions, please submit feedback on GitHub Issues. We also welcome community contributions to the code and documentation.

📄 License

MCP-Bridge is licensed under the MIT License. Please refer to the LICENSE file for details.

Thank you for using MCP-Bridge! If you have any questions, please feel free to contact us.