watch

Watch a video (URL or local path). Downloads with yt-dlp, extracts auto-scaled frames with ffmpeg, pulls the transcript from captions (or Whisper API fallback), and hands the result to Claude so it can answer questions about what's in the video.

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

High

Do not use without reviewing

Fix and improve this skill with Tessl

tessl review fix ./skills/watch/SKILL.md

Quality

Content

85%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-sequenced skill body with concrete commands, explicit validation checkpoints, and a clean single-file structure backed by verified bundled scripts; the only real weakness is repeated restatement of detail-mode caps and flag semantics across sections.

Suggestions

Consolidate the detail-mode cap table (currently restated in Step 0, 'Recommended limits', 'How to invoke', and 'Detail and frames') into one canonical table and cross-reference it from the other sections to save tokens.

Trim the repeated flag definitions in 'How to invoke' where they duplicate the Step 0 and 'Recommended limits' descriptions.

Dimension	Reasoning	Score
Conciseness	The body is almost entirely task-specific and executable with no pedagogical filler about concepts Claude knows, but detail-mode caps and flag semantics are restated across Step 0, 'Recommended limits', 'How to invoke', and 'Detail and frames', adding repetition that could be tightened — so it is mostly efficient rather than fully lean.	2 / 3
Actionability	Provides fully executable, copy-paste commands (e.g. 'python3 "${SKILL_DIR}/scripts/watch.py" "<source>"') with every flag given concrete values and worked examples, matching the score-3 executable anchor rather than the pseudocode of a 2.	3 / 3
Workflow Clarity	A clearly sequenced Step 0→5 flow with explicit validation checkpoints (setup preflight --check exit-code table, SKILL_DIR guard, error-recovery loops in failure modes) meets the clear-sequence-with-validation score-3 anchor, not the checkpoint-missing pattern of a 2.	3 / 3
Progressive Disclosure	A single, well-sectioned SKILL.md with all six bundled scripts (watch.py, download.py, frames.py, transcribe.py, whisper.py, setup.py — verified present) enumerated in one place and no nested references; per the simple-skill guideline, well-organized content with no need for external references scores 3.	3 / 3
	Total	11 / 12 Passed

Description

67%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A specific, action-rich description that clearly conveys what the skill does, but it lacks an explicit 'Use when…' trigger clause and only partially covers the natural phrasings a user would actually say.

Suggestions

Append an explicit trigger clause, e.g. 'Use when the user pastes a video URL or points at a local video file and asks what's in it, for a summary, or for a transcript.'

Add common user phrasings ('video URL', 'summarize this video', 'subtitles', 'what's in this video') to broaden trigger-term coverage.

Drop the technical 'auto-scaled frames with ffmpeg' phrasing from the description in favor of user-facing terms, keeping implementation detail in the body.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Downloads with yt-dlp', 'extracts auto-scaled frames with ffmpeg', 'pulls the transcript from captions (or Whisper API fallback)' — matching the multi-action score-3 anchor rather than the domain-plus-some-actions of a 2.	3 / 3
Completeness	The 'what' is answered thoroughly, but there is no explicit 'Use when…' trigger clause; the only 'when' is implied ('answer questions about what's in the video'), which per the guidelines caps completeness at 2.	2 / 3
Trigger Term Quality	Has relevant natural terms ('Watch a video', 'video', 'captions', 'transcript') but leans technical ('auto-scaled frames with ffmpeg') and omits common user phrasings like 'video URL' or 'summarize this video', falling short of the full coverage of the score-3 example.	2 / 3
Distinctiveness Conflict Risk	The video-watching niche via yt-dlp/ffmpeg/transcript is a clear, distinct trigger set unlikely to conflict with other skills, matching the score-3 anchor rather than the overlapping-generic of a 2.	3 / 3
	Total	10 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: bradautomates/claude-video
Path: skills/watch/SKILL.md
Commit: 83da59f

Reviewed: ~1 month ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.