MCP.so
ログイン
Y

Youtube Transcript MCP Server

@hancengiz

基本情報

カテゴリ

メディアとデザイン

トランスポート

stdio

作者

hancengiz

投稿者

Kya Huyi

Youtube Transcript MCP Server について

概要はまだありません

Youtube Transcript MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "youtube-transcript": {
      "command": "npx",
      "args": [
        "@fabriqa.ai/youtube-transcript-mcp@latest"
      ]
    }
  }
}

Youtube Transcript MCP Server のツール

ツール一覧はまだありません

この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。

上のボタンで一時的なサンドボックスでこのサーバーを実行し、ツール一覧を取得できます。

概要

What is Youtube Transcript Mcp?

Youtube Transcript Mcp is a Model Context Protocol (MCP) server that retrieves transcripts from YouTube videos for Claude Code. It allows users to extract video transcripts without manual downloading, making it useful for analyzing video content, summarizing talks, or extracting information from educational videos.

How to use Youtube Transcript Mcp?

Install globally via npm (npm install -g @fabriqa.ai/youtube-transcript-mcp) and configure it using claude mcp add --scope user youtube-transcript npx @fabriqa.ai/youtube-transcript-mcp@latest or manually in Claude configuration files. Once configured, restart Claude Code to access the get-transcript and get-transcript-languages tools.

Key features of Youtube Transcript Mcp

  • Extract full transcripts from any YouTube video with available captions.
  • Support for multiple URL formats (youtube.com, youtu.be, video ID).
  • Option to include or exclude timestamps in transcript output.
  • Request transcripts in specific languages when available.
  • Graceful error handling for videos without transcripts or invalid URLs.
  • Efficient context usage – only the transcript is returned.

Use cases of Youtube Transcript Mcp

  • Summarize key learnings from hour-long technical talks.
  • Analyze conference talks, academic lectures, and educational content without watching.
  • Get accurate quotes from video content for blog posts or articles.
  • Convert video content to searchable, readable text format.
  • Extract quotes and insights from podcast interviews and panel discussions.

よくある質問

1

What is the 25,000 token limit?

The MCP protocol has a 25,000 token response limit. Very long transcripts (60+ min with timestamps) may exceed this. Disabling timestamps typically reduces response size by 20–30%, making most videos fit within the limit.

2

Which YouTube URL formats are supported?

The tool supports `https://www.youtube.com/watch?v=VIDEO_ID`, `https://youtu.be/VIDEO_ID`, `https://m.youtube.com/watch?v=VIDEO_ID`, and just the 11-character video ID.

3

What are the runtime requirements?

The server requires Node.js installed and in the PATH. It runs as a local Node.js process communicating via stdio, using the `@modelcontextprotocol/sdk` and a custom YouTube transcript library with zero external dependencies for fetching.

4

What if a video has no transcript or language not found?

Not all YouTube videos have transcripts; some only have auto-generated captions in certain languages. Private or restricted videos cannot be accessed. Use the `get-transcript-languages` tool to check available languages.

5

How are transcripts fetched?

The server directly integrates with YouTube's Innertube API for reliable transcript access. It uses native fetch API and zero external dependencies for transcript fetching.

コメント

「メディアとデザイン」の他のコンテンツ