Content
65%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 well-structured overview skill: lean principle statements, excellent one-level-deep progressive disclosure verified against the actual bundle, and a concrete output contract. The main weaknesses are qualitative directives without executable specifics and a workflow whose sequence must be inferred from section order rather than explicit steps.
Suggestions
State the workflow as an explicit numbered sequence — check apple-design-context.md, ask only unanswered intake questions, consult the matching references/ file, then respond in the Output Format — so the flow is written rather than implied.
Inline a few executable specifics (widget size families, system update budget figures, or one worked example recommendation) so directives like 'Budget updates wisely' and 'Feature content that entices' become actionable.
Add guidance on selecting which references/ file to consult per platform, so Claude knows how to route a question about, say, watch complications vs. tvOS top shelf.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is a terse, principle-per-line overview with no explanations of concepts Claude already knows, and it correctly defers detail to references. Minor trimmable padding remains: the boilerplate 'When to Use' line ('This skill is applicable to execute the workflow or actions described in the overview') and the promotional footer add tokens without value, holding it below 5. | 4 / 5 |
Actionability | Concrete directives are present ('3-4 most common tasks', 'Deep-link on tap', 'Support multiple sizes with distinct layouts', 'Use threading and grouping') plus a five-part Output Format and intake questions, but many principles stay qualitative ('Budget updates wisely', 'Feature content that entices') with no inline specifics — no widget size families, update budget numbers, or a worked example recommendation. Not score 4 because these gaps are more than minor for an instruction-only skill. | 3 / 5 |
Workflow Clarity | Workflow components exist — an explicit context-file checkpoint ('Check for .claude/apple-design-context.md... only ask for information not already covered'), a Questions-to-Ask intake list, and an Output Format — but the overall sequence connecting them is implicit: the sections are parallel lists rather than ordered steps, and there is no verification checkpoint on the final recommendation. Not score 2 because the pre-check step and output template give a coherent rough sequence. | 3 / 5 |
Progressive Disclosure | A clear overview body pairs with a well-labeled Reference Index table describing each topic; all nine referenced paths match actual files in references/, and inspection of those files shows they are one level deep (external Apple URLs only, no nested local references). Navigation is easy and the split is appropriate — details live in references while the body stays an overview. | 5 / 5 |
Total | 15 / 20 Passed |