Conciseness

The body is dense with actionable content and never explains concepts Claude already knows (no 'what is WCAG' or 'how libraries work' filler), fitting anchor 4. It is not a 5 because 'Tool Use Is Mandatory' repeats the scan-don't-assume instruction three ways, and the five Invocation Examples (~25 lines) plus the closing '## Scope' section restate the description's triggers — padding that assumes the model needs anti-laziness hand-holding.

Actionability

Nearly every detection category ships literal, copy-paste-ready search patterns ("`.font(.system(size:` without `relativeTo:`", "`UIFont(name:` without UIFontMetrics"), executable fixes ("`.custom(\"FontName\", size: X, relativeTo: .body)`", "`UIFontMetrics(forTextStyle: .body).scaledValue(for: 20.0)`"), concrete thresholds ("44x44pt", "4.5:1 for text, 3:1 for large text"), and a full output template. The few judgment-based detections (contrast pairs, focus traversal) are inherent to static analysis and are explicitly routed to the simulator-tester agent for live validation, so anchor 5 is the closest fit.

Workflow Clarity

Five phases are explicitly sequenced (map surface → detect anti-patterns → reason about completeness → cross-reference → score) with validation checkpoints throughout: "For every grep match, use Read to verify the surrounding context before reporting", "Present this map in the output before proceeding", and "Require evidence from the Phase 1 map — don't speculate". The False Positives section acts as an error-recovery loop against over-flagging, and a Testing Checklist closes the process. This is a read-only audit, so the destructive/batch validation cap does not apply; anchor 5 matches.

Progressive Disclosure

No bundle files exist (references/, scripts/, assets/ are absent), and the skill is a self-contained agent prompt with strong internal header structure and clearly signaled cross-skill pointers ("see `axiom-tools (skills/xcui-ref.md)`", "`axiom-design (skills/typography-ref.md)`"). This fits anchor 4 ("Good structure; most content is appropriately placed; references mostly clear"). It is not a 5 — the monolithic >50-line body has no overview-to-reference split, and the Invocation Examples and '## Scope' sections restate the description; not a 3 because navigation is easy and existing references are clearly signaled, not buried.

Specificity

The description names its domains concretely ("accessibility checking, App Store submission, code review, or WCAG compliance") but contains no action verbs stating what the skill does — no "audits", "detects", or "reports". This matches anchor 2 ("Names the domain but actions are minimal or generic"): it is not entirely vague (so not 1), but it lists zero concrete actions (so not 3, which requires 1-2).

Completeness

The description is entirely a "Use when the user mentions..." clause with no statement of the skill's capabilities — the 'what' half is missing. This exactly matches anchor 2 ("only 'when' is present without 'what'"): it cannot be 3 because no 'what' is present at all, and not 1 because the 'when' clause is specific rather than extremely vague.

Trigger Term Quality

All four triggers — "accessibility checking", "App Store submission", "code review", "WCAG compliance" — are natural phrases users would say, mirroring the skill's own invocation examples ("check my app for accessibility issues", "Check if my UI follows WCAG guidelines"). A few natural terms are missing (VoiceOver, Dynamic Type, a11y, accessibility audit), which fits anchor 4 rather than 5.

Distinctiveness Conflict Risk

"accessibility checking" and "WCAG compliance" are a distinct niche, but the "code review" and "App Store submission" triggers would fire on any generic review or submission request, overlapping unrelated skills. This is more than minor overlap (so not 4), but the primary terms keep it out of the very-broad anchor 2 territory.