gemini-mcp

A minimal MCP (Model Context Protocol) server that provides access to Google's Gemini API with Google Search grounding for up-to-date information.

Features

• Single ask_gemini tool for querying Gemini models
• Google Search grounding for up-to-date information
• Returns grounding sources with responses
• Lightweight implementation using Bun

Installation

Prerequisites

• Bun runtime
• Gemini API key

Install from npm

bunx @nhosoya/gemini-mcp

Install from source

git clone https://github.com/nhosoya/gemini-mcp.git
cd gemini-mcp
bun install

Configuration

Environment Variable

Set your Gemini API key:

export GEMINI_API_KEY=your-api-key-here

Claude Code

claude mcp add gemini --scope user --env GEMINI_API_KEY=your-api-key -- bunx @nhosoya/gemini-mcp

Other MCP Clients

Example configuration for other MCP clients (e.g., Cursor):

{
  "mcpServers": {
    "gemini": {
      "command": "bunx",
      "args": ["@nhosoya/gemini-mcp"],
      "env": {
        "GEMINI_API_KEY": "your-api-key"
      }
    }
  }
}

Or if installed from source:

{
  "mcpServers": {
    "gemini": {
      "command": "bun",
      "args": ["run", "/path/to/gemini-mcp/src/index.ts"],
      "env": {
        "GEMINI_API_KEY": "your-api-key"
      }
    }
  }
}

Usage

Tool: ask_gemini

Ask Gemini a question with optional Google Search grounding.

Parameters:

Parameter	Type	Default	Description
prompt	string	(required)	The question or prompt to send to Gemini
model	string	gemini-flash-latest	Gemini model to use
grounding	boolean	true	Enable Google Search grounding

Example:

What are the latest features in Bun 1.2?

Response:

The tool returns Gemini's response along with grounding sources (when enabled):

Bun 1.2 introduces several new features...

## Sources
- [Bun 1.2 Release Notes](https://bun.sh/blog/bun-v1.2)
- [Bun Documentation](https://bun.sh/docs)

Development

# Run the server
bun run start

# Type check
bun run tsc --noEmit

License

MIT