This tool functions as a plugin or skill for Claude Code and Claude.ai, giving the AI assistant the ability to process and reason about video content. It uses yt-dlp and ffmpeg to download videos and extract frames at auto-scaled rates, pulls timestamped transcripts from free captions when available, and falls back to the Whisper API for transcription. Every frame is then read as an image by Claude, allowing the model to answer questions about visual and audio content in videos without watching them directly.