watch
Watch Skill processes videos locally by extracting scene-aware frames, running OCR on text within them, and transcribing audio via offline Whisper or captions. It builds a searchable index so follow-up questions are answered without re-processing the video. No API keys required for core features; vision and STT are optional and configurable.
Watch Skill lets you process videos with OCR, transcription, and frame indexing to answer questions about their content.
AI-generated summary based on this skill's SKILL.md
Decision gist · record as of 2026-07-12
Watch Skill lets you process videos with OCR, transcription, and frame indexing to answer questions about their content. Watch Skill processes videos locally by extracting scene-aware frames, running OCR on text within them, and transcribing audio via offline Whisper or captions. It builds a searchable index so follow-up questions are answered without re-processing the video. No API keys required for core features; vision and STT are optional and configurable.
Use it when
- Yes.
- Watch Skill extracts structured data including scene-aware frames, text via OCR, full audio transcription with timestamps, and caption data.
Verify before relying
Read SKILL.md below before installing (1 file). Open directory: indexed for reading, not audited.
Install
oxbshw/watch-skill/watch · repository language: Python
Open directory. Skills are indexed for reading, not audited. Review a skill's body before installing it.
Frequently asked questions
AI-generated answers based on this skill's SKILL.md and metadata
How do I watch videos in Claude with the watch skill?
Watch Skill processes videos locally by extracting scene-aware frames, running OCR on text within them, and transcribing audio via offline Whisper or captions. It builds a searchable index so you can ask follow-up questions about the video content without re-processing. No API keys are required for core features.
Can I extract frames from video offline without uploading?
Yes. Watch Skill processes video content offline without uploading files to cloud services. It extracts frames locally, performs OCR on detected text, and transcribes audio using offline Whisper or existing captions—all on your machine with no external API calls required for core functionality.
What does watch skill video processing extract from files?
Watch Skill extracts structured data including scene-aware frames, text via OCR, full audio transcription with timestamps, and caption data. It indexes all extracted content so you can search across multiple videos, ask follow-up questions about specific scenes or text, and retrieve results without reprocessing.
How can I search text and captions across multiple videos?
Watch Skill builds a persistent indexed database of all extracted content—frames, OCR text, transcripts, and captions—from every video you process. You can then search this index and ask follow-up questions about any video's content, with results returned instantly from the database rather than requiring re-analysis.
Is video OCR and transcription available without an API key?
Watch Skill's core features—frame extraction, indexing, and search—require no API keys. Vision and speech-to-text are optional and configurable; you can use offline Whisper for transcription or rely on existing captions, keeping your video processing completely local and private.
SKILL.md
Rendered from the published skill. Quoted content, verbatim.
/watch (Watch Skill)
You don't have a video input; this skill gives you one. It is a thin wrapper
around the watch-skill CLI — all logic lives in the engine, so this skill
works identically on every harness (Claude Code, Codex, Cursor, ...).
This is a drop-in upgrade of the classic claude-video /watch skill:
same invocation shape, plus a persistent index (ask answers follow-ups
without re-processing), OCR on frames, scene-aware sampling with perceptual
dedup, local Whisper (offline by default, no API key needed), and THE LOOP
(capture -> critique -> fix -> re-capture) for iterating on your own output.
Step 0 — Preflight (first invocation per session)
```bash watch-skill
(truncated - see the full file via the links below)
File tree — 1 file
adapters/claude-skill/skills/watch/SKILL.md
Let your AI agent find skills like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 56,283 agent skills by what they can do, searchable in plain language.
wish › “Watch and extract structured data from video files with OCR, transcription, and frame indexing”
Give your agent the search over MCP, or paste the wish link into any chat. No install? Search from any chat →
Related skills
Query any video you've already watched to get answers backed by timestamps and confidence scores, drawn from a persistent index rather than re-processing. The skill refuses unclear answers honestly and escalates automatically when uncertain, saving tokens in the process.
Watch Skill processes any video—URLs, local files, streams, recordings—by extracting frames, running OCR, and transcribing audio into a unified searchable index. Query the index directly instead of re-watching, and batch-process playlists or folders in one pass.
When a video answer goes wrong, report the correction and Watch Skill records it as a lesson stored locally. The system classifies the mistake, injects it into similar future questions, and can re-ask the original question to verify the fix worked. Check what's been learned or review token savings anytime.
Recovering from Errors guides you through diagnosing Watch Skill failures using the built-in doctor command, which identifies issues with ffmpeg, yt-dlp, video acquisition, provider authentication, and local models. Apply the stated fixes, retry your original operation, and report any persistent structured errors without looping or changing security policies.
Pull structured data from videos you've already watched: titled chapters with timestamps for long content, pinpointed bug reports with error frames and OCR text, or opening-hook analysis scored across attention, pacing, visuals, and on-screen text. All answers come straight from the persistent index with no extra model calls.
Video Memory maintains a searchable index of every video watched on your machine, preserving frames, transcripts, and extracted answers in persistent storage. Search across all indexed videos by keyword or semantic meaning, drill into specific moments, or synthesize answers from insights spanning multiple videos. Batch-import playlists or folders to build your video library quickly.
More skills watch (MIT) · watch (MIT) · claude-real-video-for-agents (MIT) · Video Watch (NOASSERTION) · Youtube Transcript (CC0-1.0) · sharing-results (MIT) · claude-real-video (MIT)