ux-audit

Audit UX patterns for quality — navigation, flows, feedback, consistency, cognitive load, context

SKILL.md

GitHub

UX Audit

Score UX patterns against the vibe-coded → engineered spectrum. Complements [ui-audit] (visual/technical) with behavioral/experiential analysis.

Usage

[ux-audit]                          # Audit the dashboard
[ux-audit] $PROJECT_ROOT/dashboard/web/components/  # Specific directory

Execution

1. Resolve target path

Default: $PROJECT_ROOT/dashboard/web/ If $ARGUMENTS provided, use that path.

2. Run the audit engine

python3 -c "
from your_project.modules.domain.ux_arbiter import run_audit
result = run_audit('TARGET_PATH')
print(result.summary)
print()
for f in result.findings:
    icon = 'PASS' if f.score == 2 else ('WARN' if f.score == 1 else 'FAIL')
    print(f'  [{icon}] {f.criterion}: {f.message}')
    if f.evidence:
        print(f'         Evidence: {f.evidence}')
print()
print(f'Files scanned: {result.files_scanned}')
"

3. Format output

UX Audit | <target> | Grade <X> (<score>/<max>)
═══════════════════════════════════════════════

## Score Breakdown
| Category        | Score | Max | Status |
|-----------------|-------|-----|--------|
| Navigation      | X     | 8   | ✓/⚠/✗  |
| Flow            | X     | 10  | ✓/⚠/✗  |
| Feedback        | X     | 8   | ✓/⚠/✗  |
| Consistency     | X     | 6   | ✓/⚠/✗  |
| Cognitive Load  | X     | 6   | ✓/⚠/✗  |
| Context         | X     | 8   | ✓/⚠/✗  |

## Failures / Warnings / Passing
(same format as [ui-audit])

## Paired Scores (if both audits run)
| Arbiter | Grade | Score |
|---------|-------|-------|
| UI      | X     | X/60  |
| UX      | X     | X/46  |
| Combined| X     | X/106 |

Categories Explained

  • Navigation: Can users find things? Breadcrumbs, sidebar, search, routing depth.
  • Flow: Can users DO things? Progressive disclosure, drill-down, pagination, confirmations.
  • Feedback: Does the system TELL users what happened? Toasts, progress, optimistic updates, undo.
  • Consistency: Is it the SAME everywhere? Naming, component reuse, data fetching patterns.
  • Cognitive Load: Is it SIMPLE? Component size, conditional complexity, information grouping.
  • Context: Does it EXPLAIN itself? Tooltips, help text, labels, onboarding.