Conciseness

The body is dense with earned content — measured costs ("~135 µs per call", "+95 MB over an 800-row scroll"), exact grep patterns, false-positive lists, and per-pattern verification rules — and mostly avoids teaching concepts Claude already knows. Not 5 because there is trimming available: the long Output Format template, Invocation Examples, and Scope sections partly restate content, and several rules are explained twice across Phases 2-4; not 3 because the padding is minor relative to genuinely non-obvious detail.

Actionability

Every detection pattern ships a verbatim grep string ("DateFormatter()", ".first(where:", "preparingThumbnail"), a verification step, and a concrete fix direction ("Use `.task` with async loading, store in @State"), and exclusions are given as literal glob paths. Not 5 because fixes are stated as one-line directions rather than the copy-paste code examples the output format itself demands ("Fix: Code example showing the fix"), leaving the agent to author each fix from scratch.

Workflow Clarity

Five phases are explicitly sequenced (map hierarchy → detect anti-patterns → reason about context → cross-reference → score), with a required intermediate artifact ("Present this map in the output before proceeding"), per-match validation ("use Read to verify the surrounding context before reporting"), a false-positive rejection list, and output limits for overflow. This matches the anchor for a clear sequence with explicit validation steps and error-recovery feedback loops; the read-verify-discard loop on every grep hit is exactly that.

Progressive Disclosure

The body is well-sectioned by phase and severity, and cross-skill references ("axiom-swiftui (skills/layout-ref.md)", "axiom-performance (skills/memory-debugging.md)") are clearly signaled in a Related section. No bundle files exist, so all content lives inline; not 5 because ~280 lines of 11 detection patterns plus context tables is a natural candidate for splitting into a patterns reference file, and the external skill references are not resolvable within this skill's bundle; not 3 because what is inline is well-organized and navigable rather than a wall of text.

Specificity

"expensive bodies, formatters, whole-collection dependencies, and missing lazy containers" names several concrete, specific items in the SwiftUI performance domain, though they are framed as issue triggers rather than explicit actions the skill performs. Not 5 because no action verbs state what the skill actually does (e.g., 'Analyzes... detects...'); not 3 because the list goes beyond 1-2 items and is quite concrete.

Completeness

The 'when' is explicit and strong ("Use when the user mentions SwiftUI performance, janky scrolling...") and the trailing clause conveys what is detected ("expensive bodies, formatters, whole-collection dependencies"), so both what and when are present. Not 5 because the 'what' is implied by the trigger list rather than stated as a distinct capability sentence; not 3 because neither half is missing or merely weakly implied.

Trigger Term Quality

"SwiftUI performance, janky scrolling, slow animations, or view update issues" are natural phrases users would say. Not 5 because common synonyms the body itself uses — "frame drops", "laggy", "hitches" — are absent from the description; not 3 because coverage of the natural vocabulary is clearly good rather than partial.

Distinctiveness Conflict Risk

"SwiftUI performance" plus concrete symptoms like "janky scrolling" carves out a clear niche with distinct triggers; no plausible conflict with generic code-review, memory, or energy skills is left unaddressed (the body even delegates memory and energy to sibling skills). Clearly matches the anchor for a clear niche with minimal conflict risk.