MCP Agent Server

🚀 MCP Proxy Server Usage Guide

MCP Agent Server is a powerful tool built on the mcp-ai-agent framework, supporting the integration of multiple AI models and tools. This document will provide a detailed introduction on how to configure, connect to, and use the MCP Proxy Server.

🚀 Quick Start

📦 Installation

Clone the Repository

Clone the MCP Agent Server repository from GitHub:

git clone https://github.com/yourusername/mcp-agent-server.git
cd mcp-agent-server

Install Dependencies

Install the project dependencies:

npm install

Start the Server

Run the following command to start the MCP Proxy Server:

node src/index.js

⚙️ Configuration

Configure MCP Agent Server

Edit the config.json file to configure the proxy server as needed:

{
  "agents": [
    {
      "name": "AI Analyzer",
      "description": "An analysis tool integrating multiple AI models",
      "tools": [
        { "type": "ai", "model": "claude-3" },
        { "type": "tool", "name": "Data Statistics" }
      ]
    }
  ],
  "serverSettings": {
    "port": 3001,
    "host": "localhost"
  }
}

Configure MCP Client

In VS Code

Edit the settings.json file and add the MCP Proxy Server:

{
  "mcp.servers": [
    {
      "name": "MCP Agent Server",
      "command": "node",
      "args": ["${fileDirname}/../mcp-agent-server/dist/index.js"]
    }
  ]
}

In Other MCP Clients

Follow the documentation of the specific client and add the following configuration:

• Command: node
• Arguments: ["/full/path/to/mcp-agent-server/dist/index.js"]

💻 Usage

Start the MCP support feature in VS Code and select the configured proxy server. Enter / to view available tools and use relevant functions as prompted.

🔧 Technical Details

Advanced Configuration

Create Custom Tools

Add custom tools in the config.json file:

{
  "agents": [
    {
      "name": "Calculator",
      "description": "A basic calculator tool",
      "tools": [
        {
          "type": "custom",
          "name": "Multiplication Operation",
          "parameters": {
            "number1": "Number 1",
            "number2": "Number 2"
          },
          "execute": "multiplyNumbers"
        }
      ]
    }
  ]
}

Use MCP Servers

• Sequential Thinking Server: Break down complex problems.
• Memory Server: Store and retrieve information.
• Brave Search Server: Perform web searches.

Troubleshooting

• Ensure that the MCP Server is configured correctly.
• For npx commands, ensure the full path is used.
• Verify that all necessary API keys are available in the environment.
• Check the configuration of the MCP Client.

Development

1. Modify TypeScript files.
2. Rebuild the project using npm run build.
3. Restart the MCP Client to load the updates.

CLI Usage

Test a Proxy

Test a specific proxy directly from the command line:

npm run test-agent -- --name="Proxy Name" --prompt="Test Prompt" --context="Optional Context"

For example, test the Brave Search proxy:

npm run test-agent -- --name="Brave Search" --prompt="What is the capital of France?" --context="Geographical information needed"

List Available Proxies

Run the following command to view all configured proxies:

npm run list-agents

📄 Conclusion

The MCP Agent Server offers powerful AI tool integration and highly customizable features, suitable for various development scenarios. With the guidance in this document, you can easily configure and efficiently use the MCP Proxy Server to enhance development efficiency.