Content
73%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-gated pipeline skill with excellent workflow sequencing and validation loops, undermined by a broken bundle reference (the missing examples/master-skeleton.html scaffold the whole build depends on) and notable repetition of the same warnings across sections. Tightening the duplicated emphasis and shipping or correcting the scaffold path would raise both conciseness and actionability.
Suggestions
Ship examples/master-skeleton.html in the bundle or fix the two references to it (step 0 and the anti-patterns table) to point at the actual scaffold path — the entire step 5 placeholder workflow depends on this file.
Dedupe the repeated warnings: state the captions-are-mandatory and use-this-scaffold rules once each (e.g., in step 5 and step 0 respectively) and let the anti-patterns table carry only a one-line pointer instead of restating full rationale.
Trim step 0's meta-commentary ("that's the single most common way this skill goes off-brand", the STOP paragraph) to a compact rule + one-line rationale; the current multi-sentence emphasis costs tokens without adding new instruction.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and imperative with no explanations of concepts Claude already knows, but key warnings are repeated three to four times: the scaffold/off-brand warning appears in the "Load first" preamble, step 0's bolded paragraph, and the anti-patterns table; the captions-are-non-optional point is stated in step 4, step 5, step 6 gate 5, and again in the anti-patterns table. This goes beyond the score-4 anchor's "minor instances of over-explanation"; it is not a 2 because there is no padding or tutorial-style filler — the content is otherwise lean, high-signal instruction. | 3 / 5 |
Actionability | Guidance is almost entirely copy-paste-ready: concrete ffmpeg/heygen/whisper CLI invocations with flags, a pinned voice ID, exact gate commands, and a concrete LINES-array paste pattern. It misses 5 because step 0's critical command `cp <SKILL_DIR>/examples/master-skeleton.html project/index.html` references a file that does not exist in the bundle (no examples/ directory ships), so the foundational copy step fails as written; still well above 3 since every other command is executable and complete. | 4 / 5 |
Workflow Clarity | The 0-6 pipeline is clearly sequenced with explicit validation checkpoints and feedback loops: the word-timings hard gate before step 5, "resolve every one" for MISMATCH warnings, the whisper forced-alignment fallback, seam-gate verify, frame spot-checks, and the hard-fail caption presence gate with a recovery path ("re-check step 5's LINES population"). This matches the score-5 anchor including error-recovery loops. | 5 / 5 |
Progressive Disclosure | Structure is good: the SKILL.md body is a router over well-signaled, one-level-deep references — references/build-spec.md, references/visualization-registry.md, references/script-voice.md, references/lexicon.json, scripts/align-captions.mjs, and assets/ — all of which exist in the bundle. It falls short of 5 because the twice-referenced `examples/master-skeleton.html` does not exist in the bundle, leaving a broken navigation path on the critical path; not a 3 since all other references are real, clearly signaled, and content is appropriately split. | 4 / 5 |
Total | 16 / 20 Passed |