home / mcp / mcp server gemini mcp server
Provides an MCP server that enables Gemini 3 models to collaborate with Claude Code through a powerful CLI and diverse tools.
Configuration
View docs{
"mcpServers": {
"rlabs-inc-gemini-mcp": {
"command": "npx",
"args": [
"-y",
"@rlabs-inc/gemini-mcp"
],
"env": {
"QUIET": "true",
"VERBOSE": "true",
"GEMINI_MODEL": "gemini-3-pro",
"GEMINI_API_KEY": "YOUR_API_KEY",
"GEMINI_PRO_MODEL": "gemini-3-pro-preview",
"GEMINI_OUTPUT_DIR": "./gemini-output",
"GEMINI_FLASH_MODEL": "gemini-3-flash-preview",
"GEMINI_IMAGE_MODEL": "gemini-3-pro-image-preview",
"GEMINI_TOOL_PRESET": "minimal",
"GEMINI_VIDEO_MODEL": "veo-2.0-generate-001",
"GEMINI_ENABLED_TOOLS": "query,search,image-gen"
}
}
}
}You can run and connect to an MCP server that lets Gemini models work with Claude Code, enabling advanced collaboration, generation, editing, and analysis workflows through a streamlined command-line interface. This server setup provides practical tools for research, image and video generation, code execution, data extraction, and real-time web interactions, all accessible from your terminal or via an MCP-enabled client.
Install or run the MCP server, then connect your MCP client to access a suite of Gemini-powered capabilities. You can perform direct Gemini queries, generate images and videos, analyze code and text, summarize content, search the web with citations, and conduct multi-turn image editing and video generation. Use predefined tool presets or enable specific tool groups to balance context usage with your tasks. The workflow is designed for development teams and researchers who want to combine Gemini’s capabilities with Claude Code’s collaboration features.
Prerequisites: ensure Node.js and Bun are available on your system. You will install or run the MCP server locally using the commands shown below.
If you encounter rate limits or connection issues, verify your API key, review server status via your MCP client, and enable verbose logging to diagnose problems. Ensure the output directory permissions are correct and that your API key has access to the required Gemini features. If 4K image generation is slow, allow extra time for rendering.
For developers, you can clone the repository, install dependencies, build, and run in development mode to test changes.
Direct queries to Gemini with thinking level control and optional complexity.
Generate images with configurable prompts, aspect ratio, and size.
Start a multi-turn image editing session for iterative refinements.
Continue refining an image based on new prompts.
Close an image editing session.
List all active image editing sessions.
Generate videos with Veo, including optional aspect ratios and negative prompts.
Poll for video generation status and download when complete.
Analyze code for quality, security, and performance.
Analyze text content for sentiment, summary, or entities.
Collaborative brainstorming to generate ideas and approaches.
Summarize content at various levels of detail.
Let Gemini write and execute Python code with common data libraries.
Real-time web search with citations and source URLs.
Get JSON responses that match a provided schema.
Convenience tool for extracting entities, facts, keywords, or sentiment.
Analyze YouTube videos, including clipping and questions.
Quick video summaries with different styles.
Analyze PDFs and documents with questions and resolution settings.
Quick PDF summarization with multiple styles.
Extract tables from documents in various formats.