Content
46%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 concise, well-sectioned router document whose only serious flaw is that everything it routes to is missing: the bundle contains no reference files, so the skill's executable substance exists nowhere. The dangling references sink actionability and progressive disclosure, and the absence of any validation step caps workflow clarity.
Suggestions
Ship the reference files the body depends on: at minimum references/setup-agents.md plus the mode-specific files it routes to, so 'load the appropriate reference file and follow its steps exactly' resolves to real content.
Add a validation checkpoint to the generate/upgrade modes (e.g., verify generated configs parse and match the detected tool list before finishing), since these modes batch-write config files across multiple AI tools.
Remove the duplicated mode descriptions between the Modes section and the wizard block — the wizard can display the mode list without restating each mode's summary twice.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence — no concept explanations, no padding. Minor redundancy: the four mode descriptions appear twice (Modes section and the wizard block) and the opening line restates the description, so not a 5. | 4 / 5 |
Actionability | The wizard dialog and mode definitions are concrete, but all executable guidance is delegated: "load the appropriate reference file and follow its steps exactly" — and references/setup-agents.md does not exist in the bundle (no references/ directory at all). The specific steps to execute are missing everywhere, matching the 2 anchor (high-level hints, missing steps); not 1 because the routing and wizard text are themselves concrete. | 2 / 5 |
Workflow Clarity | A coherent top-level sequence exists (no args → wizard → load mode reference → follow steps → closing log), but per-mode steps are absent and there are no validation checkpoints for a generate mode that writes many config files across tools — a batch-write operation, which caps workflow_clarity at 3 per the rubric. | 3 / 5 |
Progressive Disclosure | Sections are well-organized and the reference index is clearly signaled as a one-level router ("read this first. It routes to mode-specific files"), but scored against the actual bundle structure the referenced files are entirely missing — no references/, scripts/, or assets/ directories exist. The primary navigation target is dangling, so the disclosure structure fails at its main job; not 1 because the SKILL.md body itself remains sectioned and readable rather than a monolith. | 2 / 5 |
Total | 11 / 20 Passed |