| name | youtube-content-analysis |
| description | Extract and analyze YouTube video transcripts and metadata. Provides tools to fetch transcripts (with or without timestamps) and video information. |
| license | Copyright (c) 2025 https://github.com/levonk. Licensed under the GNU AGPL-3.0 License. |
YouTube Content Analysis Skill
This skill enables AI agents to interact with YouTube content by extracting transcripts and metadata. It is a TypeScript-based implementation of the YouTube Transcript MCP, designed to be lightweight and fast.
Principles
- Direct Access: Use standard YouTube transcript APIs without requiring heavy browser automation where possible.
- Structured Output: Return transcripts in formats suitable for further LLM processing (plain text or timed snippets).
- Efficiency: Use TypeScript for performance and type safety.
Tools
get_transcript: Fetches the full text transcript of a video.get_timed_transcript: Fetches transcript snippets with start times and durations.get_video_info: Retrieves video metadata (title, description, duration, etc.).
Implementation
The logic is implemented in index.mts and executed via a shebang script. It leverages youtube-transcript-api (or equivalent) and mcp-sdk.