bradautomates/claude-video
TL;DR
claude-video is a Claude Code plugin/skill that enables Claude to ingest YouTube videos by processing their frames and transcripts, allowing Claude Code agents to learn from and act on video content. Enables Claude Code to consume YouTube tutorial content natively, allowing users to create specialized AI agents from any instructional video without manual transcription or summarization.
What it actually is
- What: claude-video is a Claude Code plugin/skill that enables Claude to ingest YouTube videos by processing their frames and transcripts, allowing Claude Code agents to learn from and act on video content.
- Who built it / maintained by: bradautomates (Brad Automates)
- Status: stable
- Why it matters: Enables Claude Code to consume YouTube tutorial content natively, allowing users to create specialized AI agents from any instructional video without manual transcription or summarization.
- How it compares to alternatives:
- Gemini native YouTube understanding
- OpenAI vision APIs
- Whisper-based transcription pipelines
- yt-dlp + custom LLM workflows
- GitHub stars: 14,997 · License: MIT · Archived: no
Links
Kickstarter guide
Install the plugin in Claude Code using '/plugin marketplace add bradautomates/claude-video' or '/plugin install watch@claude-video'. Once installed, send a YouTube URL to Claude Code and it will process frames and transcript into a structured timeline. Optionally pair with a Gemini API key (free via Google AI Studio) for native YouTube understanding, then reconcile both outputs into an agent spec using a skill-creator agent.