gemini

The gemini MCP server is a TypeScript-based MCP implementation that connects to Google's Gemini API directly (no deprecated SDKs) to generate text using the Gemini model. It implements the Model Context Protocol (MCP) so it can integrate with clients like Claude Desktop, maintaining conversation context for natural interactions. The primary MCP tool exposed by this server is generate_text, which lets you generate Gemini-backed responses by sending prompts through the MCP interface. To use it, start the Node server with your Gemini API key configured, then connect your MCP client (e.g., Claude Desktop) to the gemini MCP server and issue the generate_text command with your desired parameters. The server keeps track of conversational state so subsequent prompts can build on prior context for a more coherent interaction.

Prerequisites:

• Node.js 18 or higher
• Google Gemini API key
• TypeScript tooling (build steps assumed)
• Claude Desktop (for integration testing)

1. Clone the repository

git clone https://github.com/YOUR-USERNAME/gemini-mcp-server.git
cd gemini-mcp-server

2. Install dependencies

npm install

3. Build the project (if applicable)

npm run build

4. Set up environment and run the server

• Create a local environment file or pass environment variables when starting the server:

export GEMINI_API_KEY=your_gemini_api_key_here
node path/to/dist/gemini_mcp_server.js

Note: Ensure your Gemini API key has the necessary permissions for the Gemini 2.0 Flash model. If you are integrating with Claude Desktop, configure the MCP server URL and credentials in Claude as described in the integration section below.

Tips and notes:

• The server uses direct HTTP requests to Gemini’s endpoint, so ensure outbound network access to Google’s API is allowed.
• The GEMINI_API_KEY must be kept secure; do not commit it to source control.
• If you update the server code, re-build before running to ensure the latest changes are active.
• In Claude Desktop integration, the config should point to the built dist file (e.g., dist/gemini_mcp_server.js) and include the GEMINI_API_KEY in env.
• If you encounter authentication errors, verify that the API key is correct and has the Gemini API access enabled.
• You can extend the MCP tools set by adding additional endpoints that map to Gemini API calls, but be mindful of rate limits and model availability.
• For troubleshooting, run the server in a verbose mode or inspect logs to confirm that requests are reaching Gemini and that MCP command parsing is functioning as expected.