laws-of-ux-review

Mode d’emploi

Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
Consultez le document original pour les outils, identifiants et environnements requis.
Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.

npx skills add https://github.com/keysjoao/laws-of-ux-skills --skill laws-of-ux-review

Avant de commencer

L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.

Document Skill original

Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.

Laws of UX — Review & Audit

Score existing UI code against all 30 Laws of UX and deliver prioritized, code-level fixes.

Process

Step 1: Read the Target

Read the component/page file(s) the user wants audited. If no file specified, ask which file or component to review.

Step 2: Score Each Law

Load references/ux-laws-complete.md from the parent laws-of-ux skill.

For each of the 30 laws, score:

Score	Meaning
2	Compliant — law is properly applied
1	Partial — some application but incomplete or inconsistent
0	Violation — law is broken or not addressed where it should be
—	N/A — law doesn't apply to this component type

Maximum score: 60 (all 30 laws fully compliant)

Step 3: Classify Findings

Group all non-compliant laws into severity tiers:

CRITICAL (score 0, high-impact laws): Laws that directly cause usability failures:

Fitts's Law (unreachable/tiny targets)
Doherty Threshold (no loading feedback)
Hick's Law (overwhelming choices)
Cognitive Load (too much at once)
Jakob's Law (breaks conventions)

WARNING (score 0-1, medium-impact laws): Laws that degrade experience:

Proximity, Common Region, Similarity (grouping issues)
Miller's Law, Chunking (information overload)
Goal-Gradient, Zeigarnik (missing progress)
Von Restorff (nothing stands out)

SUGGESTION (score 1, polish laws): Laws that improve quality:

Aesthetic-Usability (visual polish)
Peak-End Rule (celebration states)
Flow (minor interruptions)
Parkinson's Law (time cues)

Step 4: Generate Report

Output this exact format:

markdown

# UX Audit: [Component/Page Name]

**Score: [X]/60** | **Grade: [A/B/C/D/F]**

| Grade | Range |
|-------|-------|
| A | 50-60 |
| B | 40-49 |
| C | 30-39 |
| D | 20-29 |
| F | 0-19 |

## Critical Issues ([count])

### [Law Name] — Score: 0/2
**Problem:** [What's wrong, referencing specific lines]
**Fix:**

// Before (line XX) [current code]

// After [fixed code]


## Warnings ([count])

### [Law Name] — Score: [0-1]/2
**Problem:** [description]
**Fix:** [code change]

## Suggestions ([count])

### [Law Name] — Score: 1/2
**Improve:** [description]
**Code:** [enhancement]

## Compliant ([count])
[List of laws scoring 2/2 — no action needed]

## N/A ([count])
[List of laws not applicable to this component]

Step 5: Prioritized Action Plan

After the report, provide a numbered action plan:

markdown

## Action Plan (do in this order)

1. **[Critical fix]** — [1-line description] → [file:line]
2. **[Critical fix]** — [1-line description] → [file:line]
3. **[Warning fix]** — [1-line description] → [file:line]
...

Rules

Always read the actual code — never score based on assumptions
Every violation needs a code fix — no fix = not a real finding
Be honest about N/A — don't force-apply laws that don't fit
Reference specific lines — file.tsx:42 not "somewhere in the component"
Score consistently — same violation in two components gets same score