YouTube

You can summon Claude to summarize YouTube videos by using subtitles downloaded via yt-dlp. This MCP server connects yt-dlp to Claude.ai through the Model Context Protocol, enabling concise video summaries based on the selected subtitles. It is useful when you want quick insights from long videos without watching the entire content.

How to use

Share a YouTube video URL and, optionally, a subtitle language code when you want a specific language for the subtitles. This MCP server handles fetching the video subtitles with yt-dlp and passes them to Claude for context-aware summarization. You can ask Claude to summarize by saying something like: Summarize the YouTube video <<URL>>. If you specify a language, ensure that language subtitles exist for the video.

How to install

Prerequisites you need on your machine: Node.js and npm, plus yt-dlp installed locally.

# macOS
brew install yt-dlp

# Windows
winget install yt-dlp

Configuration and usage notes

To run this MCP server, use the following configuration. It uses npx to run the published package that connects YouTube subtitles with Claude via MCP.

{
  "mcpServers": {
    "youtube": {
      "command": "npx",
      "args": ["-y", "@kazuph/mcp-youtube"]
    }
  }
}

Example

This is a ready-to-run MCP configuration snippet that demonstrates how to start the YouTube subtitle MCP server. Use it in your MCP client to load the YouTube MCP server.

Notes

• Language codes follow ISO 639-1. If you request a language that is not available for a given video, an error will occur. - If yt-dlp is not found in your environment, install it using the steps above. - The server expects the YouTube URL to be provided by your client through the MCP interface.

Available tools

SubtitleDownloader

Downloads subtitles for YouTube videos using yt-dlp and exposes the text for downstream processing in an MCP workflow.

ClaudeContextIntegrator

Passes extracted subtitles to Claude.ai and uses the Model Context Protocol to obtain a concise, context-aware summary.