Name: YouTube Transcription Agent
Availability: InStock
Author: gilbertsahumada

What it does

This MCP server lets your AI assistant transcribe and summarize YouTube videos. It downloads audio via yt-dlp, runs it through OpenAI Whisper for transcription, and returns timestamped text or concise summaries. It also exposes an Agent-to-Agent (A2A) HTTP interface with x402 micropayments on Base Sepolia, so other AI agents can pay per request.

Tools

transcribe_video — Downloads YouTube audio and transcribes it with timestamps using the Whisper API
summarize_video — Transcribes the video and generates a summary with key points
chat — General conversation with the agent

Installation

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "youtube-transcriber": {
      "command": "npx",
      "args": ["tsx", "/path/to/src/mcp-server.ts"],
      "env": {
        "OPENAI_API_KEY": "sk-..."
      }
    }
  }
}

Requires system dependencies: ffmpeg and deno (for yt-dlp).