Gemini

This MCP server lets Claude Desktop talk to Gemini AI models through a fully supported Model Context Protocol interface. It enables real-time streaming responses, secure API key handling, and configurable model parameters, all implemented in TypeScript for robust integration with your client workflows.

How to use

You connect an MCP client to this server to access Gemini model capabilities. Start the server locally or point your MCP client at a running instance, and you can request model interactions, receive streaming results in real time, and adjust key parameters to fit your use case.

Configure your client to load the Gemini MCP server as an MCP endpoint. The Gemini server is exposed through a local stdio endpoint that you run via a helper command. You’ll provide your Gemini API key as an environment variable so keys are never exposed in the client.

Example configuration snippet for your client points to the Gemini MCP server entry named gemini and passes your API key through an environment variable. This enables secure key handling while your client communicates with the MCP server.

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

How to install

Follow these concrete steps to set up the Gemini MCP server locally and run it in development mode.

Prerequisites you need installed on your system:

Step-by-step commands you should run:

# Clone the Gemini MCP server repository
# (use the exact repository URL shown in the source content)
git clone https://github.com/aliargun/mcp-server-gemini.git
cd mcp-server-gemini

# Install dependencies
npm install

# Start the development server
npm run dev

Configuration

Configure Claude Desktop to load the Gemini MCP server as shown in the example configuration snippet. Save the snippet to your Claude desktop config file to wire the MCP server into your workflow.

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

Security

API keys are handled exclusively via environment variables. No keys are logged or stored by the server, and you should rotate keys regularly. Ensure that your deployment environment restricts access to the config files that contain sensitive values.

Troubleshooting

If you encounter connection issues, verify that the port used by the MCP client is available and that your internet connection is functioning. If API key problems persist, confirm the key is correct and has the necessary permissions.

Available tools

mcp_protocol

Full MCP protocol support including real-time streaming and command routing between the client and the Gemini model.

streaming

Real-time streaming of model responses to MCP clients as they are generated.

security

Secure handling of API keys through environment variables with no keys logged or stored.

configurable_parameters

Configurable model parameters exposed via the MCP server configuration.

implementation_ts

TypeScript implementation provides strong typing and integration with modern Node environments.