Gemini MCP Server

What is Gemini MCP Server?

A multi-context processing (MCP) server that uses the Gemini API to take screenshots of webpages and analyze them. It is designed for developers who need to capture, tag, and retrieve webpage screenshots along with AI-generated descriptions and metadata.

How to use Gemini MCP Server?

Clone the repository, create a .env file based on .env.example, run npm install, then npm run build, and start the server with npm start. Screenshots are captured by sending a POST request to /process with a URL and a prompt, and can be retrieved via GET /screenshots endpoints.

Key features of Gemini MCP Server

• Screenshot and analyze webpages using the Gemini API
• Add tags and metadata to screenshots
• Retrieve screenshots by ID or with filtered queries

Use cases of Gemini MCP Server

• Capture initial page states for visual regression testing
• Archive webpage visuals with AI-generated descriptions for auditing
• Build a searchable gallery of webpage screenshots filtered by tags or metadata

FAQ from Gemini MCP Server

What does the server do?

It takes a screenshot of a given URL, sends it to the Gemini API along with a user-provided prompt, and returns the screenshot, tags, and metadata.

What are the dependencies?

Node.js, a Gemini API key, and access to a Supabase project (for storing screenshots and data).

What API endpoints are available?

POST /process to capture and analyze a webpage, GET /screenshots/:id to retrieve a specific screenshot, and GET /screenshots with optional tag or key and value query parameters to filter results.

Where are screenshots stored?

Screenshots are stored in Supabase storage, as indicated by the download_url in the response (e.g., https://yoursupabaseproject.supabase.co/storage/v1/object/public/screenshots/...).