Content
100%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.
An exemplary instruction-only skill body: terse, battle-tested guidance with concrete CLI commands, a clear build-and-verify workflow with feedback loops, and clean splitting of detail into three real, well-signaled reference files. Every token earns its place.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and dense throughout: it states 'This skill holds only what earlier rebuilds got wrong or spent hours on' and delivers exactly that, assuming competence (never explaining ffmpeg, video formats, or basics) with no padding or time-sensitive version detail. | 5 / 5 |
Actionability | Copy-paste-ready commands with full flags ('npx hyperframes measure frames --range a,b --roi x,y,w,h', 'setsid nohup npx hyperframes render -o <out> > render.log 2>&1 &') plus worked specifics like 'frame 83 at 30 fps is 2.766666' cover the common cases. | 5 / 5 |
Workflow Clarity | The arc (auth -> measure -> rebuild layers -> spot-check -> check/render/diff -> fix -> one more round -> stop -> REVERSE.md) is clearly sequenced with explicit validation checkpoints and a feedback loop ('check, fix its errors, then render... Fix what it names... Render and diff once more, then stop') including a stop criterion. | 5 / 5 |
Progressive Disclosure | A ~31-line overview with well-signaled, one-level-deep references: references/start.md, references/people.md, and references/match.md all exist as focused files, and each is introduced with a summary of what it holds ('What decides whether drawn things match, in references/match.md: ...'). | 5 / 5 |
Total | 20 / 20 Passed |