Youtube Vision

What is Youtube Vision?

Youtube Vision is an MCP (Model Context Protocol) server that uses the Google Gemini Vision API to analyze YouTube videos. It allows users to get descriptions, summaries, answer questions, and extract key moments from videos. Designed for developers integrating YouTube video analysis into AI-powered workflows.

How to use Youtube Vision?

Install via Smithery (npx -y @smithery/cli install @minbang930/youtube-vision-mcp) or run directly with npx -y youtube-vision. Configure your MCP client (e.g., Claude Desktop) with the command and environment variables: GEMINI_API_KEY (required) and optionally GEMINI_MODEL_NAME. You can then use its four tools (ask, summarize, extract key moments, list supported models) by providing a YouTube URL and optional parameters.

Key features of Youtube Vision

• Analyzes YouTube videos using the Gemini Vision API
• Provides four distinct tools for various interactions
• Configurable Gemini model selection via environment variable
• Communicates via stdio (standard input/output)
• Lists available Gemini models that support generateContent

Use cases of Youtube Vision

• Quickly summarize a long YouTube video for digest or notes
• Extract key moments with timestamps from a tutorial or lecture
• Answer specific questions about video content without watching entirely
• Generate general descriptions for cataloging or accessibility

FAQ from Youtube Vision

What does Youtube Vision do that other tools don't?

Youtube Vision directly integrates the Google Gemini Vision API into the MCP ecosystem, allowing you to ask questions, summarize, and extract moments from YouTube videos through a simple tool interface. It does not rely on YouTube transcripts but uses visual analysis via Gemini.

What are the runtime requirements for Youtube Vision?

You need Node.js version 18 or higher and a Google Gemini API key from Google AI Studio or Google Cloud Console. The server runs locally via stdio.

Where does Youtube Vision store video data?

The README does not specify data storage. Video analysis is performed on-demand through the Gemini API, and generated content is returned to the MCP client. Users are responsible for complying with Google’s data usage policies.

Are there any known limits of Youtube Vision?

The README does not list specific rate or usage limits. It notes that model versions marked as “Experimental” or “Preview” are not permitted for production use per Gemini API terms.

What transport and authentication does Youtube Vision use?

It communicates via stdio transport and authenticates using the GEMINI_API_KEY environment variable, which must be set in the MCP client’s configuration.