Gemini MCP Server

Google Gemini API access for web search and model-driven task delegation.

Data last scanned today · Reviewed today

232 tools53PythonMIT

Overview

Gemini MCP Server acts as a bridge for AI assistants to utilize Google's Gemini models through the Model Context Protocol. It offers two primary capabilities: performing web searches that return synthesized results with citations and delegating specific prompts to Gemini 2.5 Pro or Flash models.

Our verdict

With a security score of 18/100, this server is classified as high-risk. The repository has not received a commit in over 452 days and lacks the oversight of official vendor maintenance. Given these factors, it is recommended only for users who can independently audit the code and manage the risks associated with an unmaintained, community-provided integration.

  • Supports both STDIO and streamable-http transport modes
  • Zero direct dependencies reduce supply-chain surface area
  • Uses standard MIT licensing
  • Includes documented support for Gemini 2.5 model selection
  • Last code update occurred over 452 days ago
  • Maintained by an individual contributor rather than Google
  • No official editorial or security review status

Tools

ToolDescriptionRisk
web_searchPerforms a web search using Gemini and returns synthesized results with citationslow
use_geminiDelegates a task to a specified Gemini 2.5 model (Pro or Flash).low

Compatibility

ClientLocalDockerRemoteRead-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

What authentication methods does the Gemini MCP server support?

For STDIO mode, the server uses the GEMINI_API_KEY environment variable. For HTTP mode, it requires a Bearer token provided in the Authorization header.

Which Gemini models are supported by the use_gemini tool?

The use_gemini tool allows you to delegate tasks to Gemini 2.5 Pro or Flash models. It defaults to using gemini-2.5-flash-preview-05-20 if no model is specified.

Can I deploy the Gemini MCP server to the cloud?

Yes, you can deploy the server to Google Cloud Run. The provided deployment instructions cover building the Docker image, pushing it to the Artifact Registry, and configuring the service.

How can I test the server locally?

You can use the MCP Inspector to test the server by running it in either stdio or streamable-http mode. Additionally, you can run the project's test suite using pytest, provided you have set the GEMINI_API_KEY environment variable.

Does the web_search tool support citations?

Yes, you can enable citations by setting the include_citations parameter to true. When enabled, the tool returns the search results along with relevant citation metadata.

Badge

Gemini MCP Server security score, rated on RepoAI

Maintain this server? Add the live badge to your README.