Examples and Usage Instructions

December 15, 2024 ยท View on GitHub

Claude Desktop Setup

  1. Locate the Configuration File

    Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json

  2. Add Gemini MCP Configuration

    {
  "mcpServers": {
    "gemini": {
      "command": "npx",
      "args": ["-y", "github:aliargun/mcp-server-gemini"],
      "env": {
        "GEMINI_API_KEY": "your_api_key_here"
      }
    }
  }
}

  3. Restart Claude Desktop

    • Close Claude Desktop completely
    • Relaunch the application
    • The Gemini provider should now be available

Example Interactions

Basic Usage

Claude: I can now access Gemini through the MCP connection. Would you like me to compare our responses to a question?

Human: Yes, please compare how we respond to: "Explain quantum computing in simple terms."

Claude: I'll use both my own knowledge and ask Gemini through the MCP connection...

Advanced Features

  1. Parameter Control

    {
  "method": "generate",
  "params": {
    "prompt": "Your prompt here",
    "temperature": 0.7,
    "maxTokens": 1000
  }
}

  2. Streaming Responses

    {
  "method": "generate",
  "params": {
    "prompt": "Your prompt here",
    "stream": true
  }
}

Troubleshooting Common Setup Issues

  1. Config File Not Found

    • Make sure Claude Desktop has been run at least once
    • Check the path for your operating system
    • Create the file if it doesn't exist

  2. API Key Issues

    • Get your API key from Google AI Studio
    • Ensure the key has proper permissions
    • Check for any whitespace in the key

  3. Connection Issues

    • Verify Claude Desktop is running
    • Check if port 3005 is available
    • Look for any firewall restrictions

Best Practices

  1. API Key Security

    • Never share your API key
    • Use environment variables when possible
    • Rotate keys periodically

  2. Resource Management

    • Monitor API usage
    • Implement rate limiting
    • Handle long responses appropriately

Advanced Configuration

Custom Port Configuration

{
  "mcpServers": {
    "gemini": {
      "command": "npx",
      "args": ["-y", "github:aliargun/mcp-server-gemini"],
      "env": {
        "GEMINI_API_KEY": "your_api_key_here",
        "PORT": "3006"
      }
    }
  }
}

Development Setup

{
  "mcpServers": {
    "gemini": {
      "command": "node",
      "args": ["/path/to/local/mcp-server-gemini/dist/index.js"],
      "env": {
        "GEMINI_API_KEY": "your_api_key_here",
        "DEBUG": "true"
      }
    }
  }
}

Using with Other MCP Servers

Multiple Providers Example

{
  "mcpServers": {
    "gemini": {
      "command": "npx",
      "args": ["-y", "github:aliargun/mcp-server-gemini"],
      "env": {
        "GEMINI_API_KEY": "your_gemini_key"
      }
    },
    "openai": {
      "command": "npx",
      "args": ["-y", "@mzxrai/mcp-openai@latest"],
      "env": {
        "OPENAI_API_KEY": "your_openai_key"
      }
    }
  }
}

Verification Steps

  1. Check Server Status
curl -I http://localhost:3005
  1. Test WebSocket Connection
wscat -c ws://localhost:3005
  1. Verify MCP Integration
Ask Claude: "Can you verify if the Gemini MCP connection is working?"