Content
57%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.
The body is a well-structured reference overview with an excellent progressive-disclosure design — a concise principle summary indexing 13 real, one-level-deep reference files. Its weaknesses are that guidance remains high-level (no concrete APIs or frameworks named), the operating workflow is implied rather than sequenced, and it carries known-concept restatement plus boilerplate filler sections.
Suggestions
Replace generic directives with named specifics, e.g. "Use SwiftUI's .scrollDismissesKeyboard / keyboard-avoidance modifiers or UIKit's keyboard notifications" instead of "Use keyboard-avoidance APIs", and name the recognizers (UITapGestureRecognizer, UIPanGestureRecognizer) when saying "use standard recognizers".
Make the workflow explicit and ordered: 1) read .claude/apple-design-context.md if present, 2) ask only the uncovered questions from Questions to Ask, 3) produce the Output Format deliverables — moving the description's check-first behavior into the body.
Delete the filler "When to Use" section ("This skill is applicable to execute the workflow or actions described in the overview" says nothing) and the attribution footer, and trim restatements of universally known gestures and shortcuts.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is compact and table-driven, but it restates concepts Claude already knows ("Tap to activate, swipe to scroll/navigate, pinch to zoom", "Standard shortcuts (Cmd+C/V/Z)") and ends with a filler section ("This skill is applicable to execute the workflow or actions described in the overview") plus an attribution footer. Mostly efficient with some removable padding, matching the level-3 anchor rather than level 4, which admits only minor trims. | 3 / 5 |
Actionability | As an instruction-only reference skill it gives a usable checklist ("Distinguish finger from Pencil when appropriate", "haptic feedback at detents", "Map to extended gamepad profile") and a four-part output format. But guidance stays high-level — "Use keyboard-avoidance APIs" and "Apple's built-in recognizers" without naming a single concrete API or framework — so it matches the level-3 anchor (concrete but incomplete) rather than level 4's mostly-executable guidance. | 3 / 5 |
Workflow Clarity | The intended process is inferable (check the context file, ask the four listed questions, produce the four-part output format) and the Output Format and Questions to Ask sections are well-specified, but the sequence is never explicitly ordered and the context-file check from the description is not operationalized into numbered steps. This matches the level-3 anchor (steps present but implicit) rather than level 4's clear sequence with checkpoints. | 3 / 5 |
Progressive Disclosure | The body is a genuinely lean overview (15 principles grouped by input category) plus a Reference Index table of 13 one-level-deep references, all of which were verified to exist in references/, each annotated with topic and key-content columns. This is a clean match for the level-5 anchor: clear overview, well-signaled one-level-deep references, appropriate content split, easy navigation. | 5 / 5 |
Total | 14 / 20 Passed |