← ClaudeAtlas

reviewlisted

Use when the user asks for a broad, comprehensive, general, or end-to-end UX review of a feature, screen, flow, app, or implementation — across multiple dimensions (usability, accessibility, psychology, product, trust) rather than one specific angle. If the request already narrows to one angle (e.g. "check accessibility", "is this manipulative"), prefer the matching domain skill instead.
DrSmile444/ux-crux · ★ 0 · AI & Automation · score 63
Install: claude install-skill DrSmile444/ux-crux
Review the user-facing product, feature, flow, screen, or implementation described or shown, across every UX dimension relevant to the evidence provided. Never invent behavior that cannot be observed — see `../../shared/evidence-model.md`. ## Procedure 1. **Determine scope.** Read `references/review-model.md`'s lens-selection procedure. Decide which of usability, psychology, accessibility, product, and trust apply, and what evidence is actually available (screenshot, description, running build, code). 2. **State the scope before findings.** Name which lenses you applied and why, and which you could not apply due to the evidence type. 3. **Evaluate each applicable lens.** This package bundles every domain's rule content under `domains/<domain>/` (generated from that domain's own skill, so it is always present here regardless of whether the domain skills are also installed). Load only the `domains/<lens>/...` files you need for the lenses in scope — do not pull in all five domains' full reference material for a narrowly-scoped review. For example: `domains/usability/core.md` and `domains/usability/mobile.md`, `domains/psychology/*.md`, `domains/accessibility/core.md` and `domains/accessibility/mobile.md`, `domains/product/core.md`, `domains/trust/core.md` and `domains/trust/mobile.md`. 4. **Tag every finding** with an evidence status (`../../shared/evidence-model.md`), a severity (`../../shared/severity-model.md`), and a confidence level. 5. **Synthesize.** Merge overlapping