Wraps Google's Gemini models as MCP tools inside Claude Desktop. The real utility isn't just chat, it's the stuff Gemini does natively: grounded search that pulls live Google results before answering, Imagen for image generation with current data lookup, Veo for video synthesis, and SVG diagram rendering that outputs clean vector code you can commit straight to a repo. Deep research runs multi-iteration searches then synthesizes a full report. Image editing maintains conversational context across turns via thought signatures. Thirteen tools total, all stdio transport. Works via npx with just an API key from Google AI Studio. Free tier covers most development use.
claude mcp add --transport stdio io.github.houtini-ai-gemini uvx gemini