UX Audit
Core Insight: Users don't read manuals. If it's not obvious, it's broken.
THE EXACT PROMPT
Conduct a UX audit of this [application/CLI/interface]:
1. Nielsen's 10 Usability Heuristics
2. Accessibility basics (keyboard, contrast, screen reader)
3. User flow analysis (happy path + error paths)
4. Cognitive load assessment
Produce prioritized report:
- Critical (blocking)
- Important (degraded experience)
- Suggestions (polish)
Include file:line references and concrete fixes.
Quick Check (5 min)
| Area | Status | Notes |
|------|--------|-------|
| Navigation | 🟢/🟡/🔴 | |
| Forms | 🟢/🟡/🔴 | |
| Errors | 🟢/🟡/🔴 | |
| Accessibility | 🟢/🟡/🔴 | |
| Mobile/Responsive | 🟢/🟡/🔴 | |
Nielsen's 10 Heuristics (Quick Reference)
| # |
Heuristic |
Key Question |
| 1 |
Visibility |
Does user know what's happening? |
| 2 |
Real World |
Familiar language & concepts? |
| 3 |
Control |
Can user undo/escape? |
| 4 |
Consistency |
Same action = same result? |
| 5 |
Error Prevention |
Prevents mistakes before they happen? |
| 6 |
Recognition |
Options visible vs. memorized? |
| 7 |
Flexibility |
Shortcuts for experts? |
| 8 |
Minimal |
Visual clutter removed? |
| 9 |
Error Help |
Error messages helpful? |
| 10 |
Documentation |
Help available when needed? |
Deep dive: HEURISTICS.md
CLI-Specific Heuristics
| Heuristic |
CLI Application |
| Visibility |
--verbose, progress bars, spinners |
| Real world |
Standard flags (--help, --version) |
| Control |
Ctrl+C works, --dry-run available |
| Consistency |
-v short, --verbose long |
| Error prevention |
Confirmation prompts, --force |
| Recognition |
--help shows all options |
| Flexibility |
Short + long flags, piping |
| Minimal |
Clean default, verbose optional |
| Error help |
Actionable messages, suggested fixes |
| Documentation |
--help, man pages, examples |
Generate checklist: ./scripts/generate-checklist.py cli
Accessibility Quick Check
The Big 4
Deep dive: A11Y.md
Output Template
# UX Audit: [Project]
## Summary
**Score:** X/10 | **Critical:** N | **Important:** N | **Suggestions:** N
## Critical Issues (Must Fix)
### 1. [Title]
**Heuristic:** #N - [Name]
**Location:** `file:line`
**Problem:** [What's wrong]
**Fix:** [Specific solution]
## Important Issues (Should Fix)
[Same format]
## Suggestions
- [ ] [Item]
## Heuristic Scores
| Heuristic | Score |
|-----------|-------|
| 1. Visibility | X/10 |
...
Full template: EXAMPLES.md
Audit Depth
| Type |
Time |
When |
| Quick Scan |
5-10 min |
PR review, spot check |
| Standard |
30-60 min |
Pre-release, feature review |
| Deep Audit |
2+ hr |
Major redesign, compliance |
Anti-Patterns
| Don't |
Do |
| "It's intuitive" |
Test with fresh eyes |
| Ignore error paths |
Audit happy AND error flows |
| Skip accessibility |
Check the Big 4 minimum |
| Vague issues |
Specific file:line + fix |
| All issues equal |
Prioritize by impact |
Integration
With Hooks
Auto-run a11y check after UI edits:
{
"hooks": {
"PostToolUse": [{
"matcher": "Edit",
"hooks": [{
"type": "command",
"command": "if echo \"$TOOL_INPUT_PATH\" | grep -qE '\\.(tsx|jsx|html)$'; then npx axe \"$TOOL_INPUT_PATH\" 2>/dev/null || true; fi"
}]
}]
}
}
With Other Skills
| Combine with... |
For... |
| multi-model-triangulation |
Multiple UX perspectives |
| multi-pass-bug-hunting |
Include UX issues in sweeps |
| ui-polish |
After fixing critical UX issues |
References
| Topic |
File |
| Full heuristic details |
HEURISTICS.md |
| Accessibility deep dive |
A11Y.md |
| Example audits |
EXAMPLES.md |
Scripts
| Script |
Purpose |
scripts/generate-checklist.py |
Generate project-specific checklist |
./scripts/generate-checklist.py web # Web app
./scripts/generate-checklist.py cli # CLI tool
./scripts/generate-checklist.py mobile # Mobile app
./scripts/generate-checklist.py api # API/DX
lev-os