watch
Watch lets you analyze any video—from YouTube, local files, or other sources—by automatically pulling frames and transcripts so Claude can answer your questions about the content. It handles caption extraction, falls back to Whisper when needed, and scales frame detail to match your needs.
Watch extracts video frames and transcripts so Claude can answer questions about what's in the video.
AI-generated summary based on this skill's SKILL.md
Install
bradautomates/claude-video/watch · repository language: Python
git clone https://github.com/bradautomates/claude-video
cp -r claude-video/skills/watch ~/.claude/skills/watchnpx skillfed install bradautomates/claude-video/watchFrequently asked questions
AI-generated answers based on this skill's SKILL.md and metadata
How do I watch a YouTube video and answer questions about it?
Watch lets you submit a YouTube URL along with your question, and it automatically extracts frames and generates a transcript. Claude then analyzes both the visual and spoken content to answer your specific questions about what happens in the video.
What video sources does Watch support for analysis?
Watch processes local video files (MP4, MOV, WebM) and URLs from major platforms including YouTube, Vimeo, TikTok, and Twitch. Whether you're working with a file on your computer or a link to online content, Watch can extract frames and transcripts for AI analysis.
How does Watch extract transcripts from videos?
Watch first attempts to extract existing captions from the video. If captions aren't available, it automatically falls back to Whisper for audio transcription. This dual approach ensures you get a readable transcript regardless of whether the video has built-in captions.
Can I focus Watch on a specific section of a video?
Yes. Watch supports time-bound analysis where you can specify a particular section. When you do, it uses denser frame sampling in that region and returns only the transcript for that timeframe, letting you zoom in on key moments without processing the entire video.
What can Watch do with the frames and transcript it extracts?
Watch combines visual frames and spoken content so Claude can summarize what happens, answer detailed questions about specific scenes, identify key moments, and provide comprehensive video understanding. You get both visual and textual evidence backing every answer.
SKILL.md
rendered from the published skill — quoted content, verbatim
/watch
You don't have a video input; this skill gives you one. A Python script gets captions first, optionally downloads the video, extracts frames as JPEGs (scene-aware, or fast keyframes at efficient detail), gets a timestamped transcript (native captions first, then Whisper API as fallback), and prints frame paths. You then Read each frame path to see the images and combine them with the transcript to answer the user.
Resolve SKILL_DIR (do this before any command)
Every python3 ... command below runs a bundled script under SKILL_DIR/scripts/. Set SKILL_DIR to the **absolute path of the directory containing THIS
(truncated - see the full file via the links below)
Read as markdown · JSON record · Browse the source repository
File tree — 10 files
skills/watch/.skillignore
skills/watch/SKILL.md
skills/watch/scripts/build-skill.sh
skills/watch/scripts/config.py
skills/watch/scripts/download.py
skills/watch/scripts/frames.py
skills/watch/scripts/setup.py
skills/watch/scripts/transcribe.py
skills/watch/scripts/watch.py
skills/watch/scripts/whisper.py