YouTube Transcript MCP Server

An MCP (Model Context Protocol) server for YouTube transcript retrieval, search, and metadata extraction — built with Cheerio (no Puppeteer/Playwright).

Tools

get_transcript

Fetch captions/transcript from a YouTube video.

Parameters:

Name	Type	Required	Description
video_url	string	✅	Full YouTube video URL (e.g. https://www.youtube.com/watch?v=dQw4w9WgXcQ)

Output: Timestamped transcript with language info.

search_videos

Search YouTube by keyword.

Parameters:

Name	Type	Required	Description
query	string	✅	Search keyword or phrase
max_results	number	❌	Maximum results (default: 10, max: 50)

Output: Titles, URLs, descriptions, and view counts.

get_video_info

Get detailed video metadata.

Parameters:

Name	Type	Required	Description
video_url	string	✅	Full YouTube video URL

Output: Title, author, duration, views, publish date, category, description.

Usage

With an MCP Client (Claude Desktop, etc.)

Add to your MCP client configuration:

{
  "mcpServers": {
    "youtube-transcript": {
      "command": "node",
      "args": ["path/to/mcp-youtube-transcript/dist/index.js"]
    }
  }
}

Direct invocation (stdio)

# List tools
echo '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | node dist/index.js

# Get transcript
echo '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"get_transcript","arguments":{"video_url":"https://www.youtube.com/watch?v=dQw4w9WgXcQ"}}}' | node dist/index.js

# Search videos
echo '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"search_videos","arguments":{"query":"MCP server tutorial","max_results":5}}}' | node dist/index.js

# Get video info
echo '{"jsonrpc":"2.0","id":4,"method":"tools/call","params":{"name":"get_video_info","arguments":{"video_url":"https://www.youtube.com/watch?v=dQw4w9WgXcQ"}}}' | node dist/index.js

Deployment to Apify

Prerequisites

• Apify CLI installed
• Apify account logged in (apify login)

Push to Apify

$apify push

The actor will be published at https://console.apify.com/actors/reverberant_equality/mcp-youtube-transcript.

Run on Apify

$apify run

Or via the Apify Console web interface.

Local Development

# Install dependencies
npm install

# Build TypeScript
npm run build

# Start the server
npm start

Architecture

• TypeScript — fully typed, compiled with tsc
• Express-based MCP SDK — uses @modelcontextprotocol/sdk with stdio transport
• Cheerio — lightweight HTML/XML parsing for scraping YouTube pages (no headless browser needed)
• Axios — HTTP client for fetching pages and caption XML
• Docker — multi-stage build for minimal production image

How It Works

The server scrapes YouTube's public pages using Cheerio:

1. Transcripts — Fetches the video page, extracts ytInitialPlayerResponse JSON, finds caption tracks, fetches the XML transcript, and formats it with timestamps.
2. Search — Fetches YouTube search results page, extracts ytInitialData JSON, and parses video renderers.
3. Info — Uses the same ytInitialPlayerResponse to extract detailed metadata.

No API keys are required — all data is obtained from publicly available YouTube pages.

License

MIT