watch-video - Watch and summarize videos via frames and transcript
Downloads videos, extracts JPEG frames and timestamped transcripts, and aligns visual and audio content to generate structured markdown notes.
Tags
Updated: 2026-09-19Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Extract JPEG frames with ffmpeg
- Fetch captions or transcribe audio
- Align frames with timestamped text
- Generate structured markdown video notes
- Delete temporary working directories
Inputs
- Explicit /watch-video command
- Video URL or local path
- Optional time range parameters
- Optional Whisper API key
Outputs
- Markdown report on stdout
- Structured markdown notes file
- Deleted temporary directory
Requirements
- Python 3.9 or higher
- ffmpeg and ffprobe on PATH
- yt-dlp on PATH
- Optional GROQ_API_KEY or OPENAI_API_KEY