YouTube MCP Server

A Model Context Protocol (MCP) server that provides YouTube search and transcript generation capabilities.

Features

• YouTube Search: Search for videos with filters for duration, sorting, and publication date
• Transcript Generation: Extract transcripts/captions from YouTube videos using RapidAPI with timestamped segments

Installation

1. Clone this repository
2. Install dependencies:

   npm install

3. Build the project:

   npm run build

Usage

Starting the Server

You must provide a YouTube API key to use this server. You can obtain one from the Google Cloud Console.

For transcript extraction, you also need a RapidAPI key for the YouTube Transcript Extractor API. You can obtain one from RapidAPI.

# Using command line arguments
node dist/index.js --youtube-api-key YOUR_YOUTUBE_API_KEY --rapidapi-key YOUR_RAPIDAPI_KEY

# Using environment variables
export YOUTUBE_API_KEY=YOUR_YOUTUBE_API_KEY
export RAPIDAPI_KEY=YOUR_RAPIDAPI_KEY
node dist/index.js

# Optional: specify custom RapidAPI host (default: youtube-transcript-extractor1.p.rapidapi.com)
export RAPIDAPI_HOST=custom-host.p.rapidapi.com

Available Tools

1. youtube_search

Search for YouTube videos based on a query.

Parameters:

• query (required): The search query
• max_results: Maximum number of results (1-50, default: 10)
• order: Sort order - "relevance", "date", "rating", "viewCount", "title" (default: "relevance")
• video_duration: Filter by duration - "any", "short", "medium", "long" (default: "any")
• published_after: Filter videos published after this date (ISO 8601 format)

Example Response:

{
  "query": "machine learning tutorial",
  "total_results": 10,
  "results": [
    {
      "position": 1,
      "videoId": "dQw4w9WgXcQ",
      "title": "Introduction to Machine Learning",
      "description": "Learn the basics of machine learning...",
      "channelTitle": "Tech Channel",
      "videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
      "viewCount": "1234567",
      "duration": "10m 30s"
    }
  ]
}

2. youtube_transcript

Get the transcript/captions for a YouTube video using RapidAPI.

Note: Requires RapidAPI key to be configured.

Parameters:

• video_id (required): YouTube video ID or URL
• language: Preferred language code (currently not used by RapidAPI endpoint)

Example Response:

{
  "videoId": "dQw4w9WgXcQ",
  "title": "Video Title",
  "channel": "Channel Name",
  "language": "en",
  "segments": [
    {
      "text": "Hello and welcome...",
      "timestamp": "00:00",
      "start": 0.0
    }
  ],
  "fullText": "Hello and welcome...",
  "hasTimestamps": true
}

Development

# Run in development mode
npm run dev

# Build the project
npm run build

# Test with MCP inspector
npm test

License

MIT