Gemini MCP Server
An MCP (Model Context Protocol) server that provides access to Google Gemini AI models.
Quick Start
- Install dependencies:
npm install
- Create a
.envfile with your Gemini API key:
GEMINI_API_KEY=your_api_key_here
- Start the server:
npm run dev
The server will run at http://localhost:3333/mcp
Available Tool
gemini.generateText
Generate text using Google Gemini models.
Parameters:
prompt(string, required): The text promptmodel(string, optional): Gemini model to use (default:gemini-2.5-pro)temperature(number, optional): Temperature for generation, 0-2 (default: 1)
Returns:
text: Generated text responsemodel: Model usedtemperature: Temperature setting used
Usage Example
import { Client as McpClient } from '@modelcontextprotocol/sdk/client/index.js';
import { StreamableHTTPClientTransport } from '@modelcontextprotocol/sdk/client/streamableHttp.js';
const transport = new StreamableHTTPClientTransport(
new URL('http://localhost:3333/mcp')
);
const client = new McpClient(
{ name: 'my-client', version: '1.0.0' },
{ capabilities: {} }
);
await client.connect(transport);
const result = await client.callTool({
name: 'gemini.generateText',
arguments: {
prompt: 'Explain AI in simple terms',
model: 'gemini-2.5-pro',
temperature: 0.7
}
});
console.log(result);
await client.close();
Testing
Run the included test client (requires server to be running):
npm test
Configuration
Environment variables:
GEMINI_API_KEY(required): Your Google Gemini API keyPORT(optional): Server port (default: 3333)
Deployment
Deploy to Google Cloud Run with automated scripts:
# 0. Prerequisites (one-time, via GCP Console)
# - Create GCP project
# - Link billing account
# - Run: gcloud auth login
# 1. Setup GCP (one-time)
./bin/setup-gcp.sh \
--project-id=my-project \
--github-user=your-username \
--github-repo=gemini-mcp
# 2. Set GitHub Secrets (follow script output)
# 3. Deploy
git push origin main
For detailed deployment instructions, see DEPLOYMENT.md.
License
ISC