review-ux-ui
À propos
Cette compétence réalise des audits UX/UI complets en appliquant les heuristiques de Nielsen, les normes d'accessibilité WCAG 2.1 et l'analyse des parcours utilisateurs. Elle est conçue pour permettre aux développeurs d'auditer les interfaces en matière de facilité d'utilisation, d'accessibilité et de charge cognitive avant leur publication. Utilisez-la pour des évaluations heuristiques, des contrôles de conformité d'accessibilité et des analyses d'utilisabilité des formulaires.
Installation rapide
Claude Code
Recommandénpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/review-ux-uiCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Review UX/UI
Evaluate UX and UI for usability, accessibility, and effectiveness.
When to Use
- Usability review before release
- Accessibility audit (WCAG 2.1 AA or AAA)
- User flow evaluation for efficiency and error prevention
- Form review for usability and conversion
- Heuristic evaluation of an existing interface
- Cognitive load and information architecture assessment
Inputs
- Required: Application to review (URL, prototype, or source)
- Required: Target user (roles, technical proficiency, context)
- Optional: User research (interviews, surveys, analytics)
- Optional: WCAG target (A, AA, or AAA)
- Optional: Specific user flows to evaluate
- Optional: Assistive tech to test with (screen reader, switch access)
Procedure
Step 1: Heuristic Evaluation (Nielsen's 10 Heuristics)
Evaluate against each heuristic:
| # | Heuristic | Key Question | Rating |
|---|---|---|---|
| 1 | Visibility of system status | Does the system always inform users about what is happening? | |
| 2 | Match between system and real world | Does the system use familiar language and concepts? | |
| 3 | User control and freedom | Can users easily undo, redo, or exit unwanted states? | |
| 4 | Consistency and standards | Do similar elements behave the same way throughout? | |
| 5 | Error prevention | Does the design prevent errors before they occur? | |
| 6 | Recognition rather than recall | Are options, actions, and information visible or easily retrievable? | |
| 7 | Flexibility and efficiency of use | Are there shortcuts for experienced users without confusing novices? | |
| 8 | Aesthetic and minimalist design | Does every element serve a purpose? Is there unnecessary clutter? | |
| 9 | Help users recognize, diagnose, and recover from errors | Are error messages clear, specific, and constructive? | |
| 10 | Help and documentation | Is help available and easy to find when needed? |
Rate violation severity:
| Severity | Description |
|---|---|
| 0 | Not a usability problem |
| 1 | Cosmetic — fix if time allows |
| 2 | Minor — low priority |
| 3 | Major — high priority |
| 4 | Catastrophic — must fix before release |
## Heuristic Evaluation Findings
| # | Heuristic | Severity | Finding | Location |
|---|-----------|----------|---------|----------|
| 1 | System status | 3 | No loading indicator during data fetch — users click repeatedly | Dashboard page |
| 3 | User control | 2 | No undo for item deletion — only a confirmation dialog | Item list |
| 5 | Error prevention | 3 | Date field accepts invalid dates (Feb 30) | Booking form |
| 9 | Error recovery | 4 | Form submission error clears all fields | Registration |
Got: All 10 heuristics evaluated with findings and severity ratings. If fail: If short on time, focus on heuristics 1, 3, 5, and 9.
Step 2: Accessibility Audit (WCAG 2.1)
Perceivable
- 1.1.1 Non-text content: Images have alt text (decorative use
alt="") - 1.3.1 Info and relationships: Semantic HTML (headings, lists, tables, landmarks)
- 1.3.2 Meaningful sequence: DOM order matches visual order
- 1.4.1 Use of colour: Colour is not the only means of conveying information
- 1.4.3 Contrast: Text contrast ≥ 4.5:1 (normal), ≥ 3:1 (large)
- 1.4.4 Resize text: Text resizes to 200% without loss of function
- 1.4.11 Non-text contrast: UI components and graphics ≥ 3:1 contrast
- 1.4.12 Text spacing: Works with line height 1.5x, letter spacing 0.12em, word spacing 0.16em
Operable
- 2.1.1 Keyboard: All functionality operable via keyboard
- 2.1.2 No keyboard trap: Focus never trapped
- 2.4.1 Skip links: Skip navigation link present
- 2.4.3 Focus order: Tab order follows logical sequence
- 2.4.7 Focus visible: Keyboard focus indicator visible
- 2.4.11 Focus not obscured: Focused element not hidden behind sticky headers/overlays
- 2.5.5 Target size: Targets ≥ 24x24px (44x44px on touch)
Understandable
- 3.1.1 Language of page:
langattribute set on<html> - 3.2.1 On focus: Focus does not trigger unexpected changes
- 3.2.2 On input: Input does not trigger unexpected changes without warning
- 3.3.1 Error identification: Errors clearly described in text
- 3.3.2 Labels or instructions: Form inputs have visible labels
- 3.3.3 Error suggestion: Error messages suggest fixes
Robust
- 4.1.1 Parsing: HTML valid (no duplicate IDs, proper nesting)
- 4.1.2 Name, role, value: Custom components have ARIA roles
- 4.1.3 Status messages: Dynamic content changes announced to screen readers
Got: WCAG 2.1 AA criteria checked with pass/fail per criterion. If fail: Use axe-core or Lighthouse for initial scan, then manual testing for criteria needing human judgement.
Step 3: Keyboard and Screen Reader Audit
Keyboard Navigation Test
Using only Tab, Shift+Tab, Enter, Space, Arrow keys, Escape:
## Keyboard Navigation Audit
| Task | Completable? | Issues |
|------|-------------|--------|
| Navigate to main content | Yes — skip link works | None |
| Open dropdown menu | Yes | Arrow keys don't work within menu |
| Submit a form | Yes | Tab order skips the submit button |
| Close a modal | No | Escape doesn't close, no visible close button in tab order |
| Use date picker | No | Custom date picker not keyboard accessible |
Screen Reader Test
Test with NVDA (Windows), VoiceOver (macOS/iOS), or TalkBack (Android):
## Screen Reader Audit
| Element | Announced As | Expected | Issue |
|---------|-------------|----------|-------|
| Logo link | "link, image" | "Home, link" | Missing alt text on logo |
| Search input | "edit, search" | "Search products, edit" | Missing label association |
| Nav menu | "navigation, main" | Correct | None |
| Error message | (not announced) | "Error: email is required" | Missing live region |
| Loading spinner | (not announced) | "Loading, please wait" | Missing aria-live or role="status" |
Got: Task flows tested with keyboard-only and screen reader. If fail: If screen reader unavailable, inspect ARIA attributes and semantic HTML as proxy.
Step 4: Analyse User Flows
Map and evaluate key user flows:
## User Flow: Complete a Purchase
### Steps
1. Browse products → 2. View product → 3. Add to cart → 4. View cart →
5. Enter shipping → 6. Enter payment → 7. Review order → 8. Confirm
### Assessment
| Step | Friction | Severity | Notes |
|------|---------|----------|-------|
| 1→2 | Low | - | Clear product cards |
| 2→3 | Medium | 2 | "Add to cart" button below the fold on mobile |
| 3→4 | Low | - | Cart icon updates with count |
| 4→5 | High | 3 | Must create account — no guest checkout |
| 5→6 | Low | - | Address autocomplete works well |
| 6→7 | Medium | 2 | Card number field doesn't auto-format |
| 7→8 | Low | - | Clear order summary |
### Flow Efficiency
- **Steps**: 8 (acceptable for e-commerce)
- **Required fields**: 14 (could reduce with address autocomplete + saved payment)
- **Decision points**: 2 (size selection, shipping method)
- **Potential drop-off points**: Step 4→5 (forced account creation)
Got: Critical user flows mapped with friction points rated. If fail: Without analytics, assess flows by task complexity and step count.
Step 5: Assess Cognitive Load
- Information density: Amount of information per screen appropriate?
- Progressive disclosure: Complex information revealed gradually?
- Chunking: Related items grouped visually (Gestalt principles)?
- Recognition over recall: Users see options rather than remember them?
- Consistent patterns: Similar tasks use similar interaction patterns?
- Decision fatigue: Too many choices at once? (Hick's law)
- Working memory: Users need to remember information across steps?
Got: Cognitive load assessed with overload or underload areas identified. If fail: Use the "squint test" — squint at the screen and check if structure and hierarchy remain apparent.
Step 6: Review Form Usability
For each form:
- Labels: Every input has a visible, associated label
- Placeholder text: Used for examples only, not as labels
- Input types: Correct HTML input types (email, tel, number, date)
- Validation timing: Errors shown on blur or submit, not every keystroke
- Error messages: Specific ("Email must include @") not generic ("Invalid input")
- Required fields: Clearly marked
- Field grouping: Related fields visually grouped
- Autocomplete:
autocompleteattributes for standard fields (name, email, address, cc-number) - Tab order: Logical flow matching visual layout
- Multi-step forms: Progress indicator shows current and total steps
- Persistence: Form data preserved if user navigates away and returns
Got: Each form assessed against the checklist with issues documented. If fail: With many forms, prioritize highest-traffic ones (registration, checkout, contact).
Step 7: Write the UX/UI Review
## UX/UI Review Report
### Executive Summary
[2-3 sentences: overall usability, most critical issues, strongest aspects]
### Heuristic Evaluation Summary
| Heuristic | Severity | Key Finding |
|-----------|----------|-------------|
[Summary table from Step 1]
### Accessibility Compliance
- **Target**: WCAG 2.1 AA
- **Status**: [X of Y criteria pass]
- **Critical failures**: [List]
### User Flow Analysis
[Key friction points with severity and recommendations]
### Top 5 Improvements (Prioritised)
1. **[Issue]** — Severity: [N] — [Specific recommendation]
2. ...
### What Works Well
1. [Specific positive observation]
2. ...
Got: Review provides prioritised, actionable recommendations with severity ratings. If fail: With too many issues, categorise into "must fix" (severity 3-4) and "should fix" (severity 1-2).
Validation
- All 10 Nielsen heuristics evaluated with severity ratings
- WCAG 2.1 criteria checked (at minimum: 1.1.1, 1.4.3, 2.1.1, 2.4.7, 3.3.1, 4.1.2)
- Keyboard navigation tested for key user flows
- Screen reader tested (or ARIA/semantic HTML reviewed as proxy)
- At least one critical user flow analysed for friction
- Cognitive load assessed
- Form usability evaluated
- Findings prioritised by severity with actionable recommendations
Pitfalls
- Confusing UX with visual design: UX is how it works; visual is how it looks. Beautiful interfaces can have terrible UX. Evaluate both but distinguish them.
- Testing only the happy path: Error states, empty states, loading states, and edge cases are where UX problems hide.
- Ignoring real devices: Browser dev tools responsive mode is a proxy. Real device testing catches touch, performance, and viewport issues.
- Accessibility as an afterthought: Accessibility issues found late are expensive to fix. Evaluate early and continuously.
- Personal preference as UX feedback: "I would prefer..." is not UX feedback. Cite heuristics, research, or established patterns.
Related Skills
review-web-design— visual design review (layout, typography, colour)scaffold-nextjs-app— Next.js application scaffoldingsetup-tailwind-typescript— Tailwind CSS for design system
Dépôt GitHub
Compétences associées
executing-plans
DesignUtilisez la compétence executing-plans lorsque vous disposez d'un plan de mise en œuvre complet à exécuter par lots contrôlés avec des points de contrôle de revue. Elle charge et examine le plan de manière critique, puis exécute les tâches par petits lots (3 tâches par défaut) tout en rapportant la progression entre chaque lot pour une revue par l'architecte. Cela garantit une mise en œuvre systématique avec des points de contrôle de qualité intégrés.
requesting-code-review
DesignCette compétence délègue un sous-agent réviseur de code pour analyser les modifications apportées au code par rapport aux exigences avant de poursuivre. Elle doit être utilisée après avoir terminé des tâches, implémenté des fonctionnalités majeures, ou avant une fusion vers la branche principale. La revue aide à détecter précocement les problèmes en comparant l'implémentation actuelle avec le plan initial.
connect-mcp-server
DesignCette compétence fournit un guide complet permettant aux développeurs de connecter des serveurs MCP à Claude Code via les transports HTTP, stdio ou SSE. Elle couvre l'installation, la configuration, l'authentification et la sécurité pour intégrer des services externes tels que GitHub, Notion et des API personnalisées. Utilisez-la lors de la configuration d'intégrations MCP, de la configuration d'outils externes ou du travail avec le Protocole de Contexte de Modèle de Claude.
web-cli-teleport
DesignCette compétence aide les développeurs à choisir entre les interfaces Web et CLI de Claude Code en fonction de l'analyse des tâches, puis permet une téléportation transparente des sessions entre ces environnements. Elle optimise le flux de travail en gérant l'état et le contexte de la session lors du passage entre le web, la CLI ou le mobile. Utilisez-la pour des projets complexes nécessitant différents outils à diverses étapes.