ui-audit — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited ui-audit (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
This skill audits a project's visual UI quality — the "does it look right and feel polished?" layer. It checks layout hierarchy, typography, spacing, color use, Gestalt grouping, interaction affordances, and pattern consistency. This sits between a11y (can people use it?) and UX (do people want to use it?).
Invoke this skill when a visual UI audit is requested. It should scan the project automatically without assuming a specific slash-command system.
This skill must work across Codex, Claude, and other SKILL.md-compatible agents. Do not assume slash commands, Claude-only memory files, or a specific frontend framework. Read the repository's project instructions and follow its existing build, styling, routing, and component conventions.
Difference from `ux-audit`:
ux-audit checks behavioral outcomes — does the flow cause drop-off? is the user confused?ui-audit checks visual implementation — is spacing consistent? is typography hierarchy clear? do controls look clickable?This is what a designer would critique in a visual review — pixel-level, component-level quality.
Before auditing, read:
AGENTS.md, CLAUDE.md, README.md, .cursor/rules, .windsurfrules, or equivalent project docs for project context and design systemsrc/index.css or equivalent for CSS custom properties, font imports, color tokensIdentify:
Go through each category below. For every issue found, note the file, the specific principle violated, and the recommended fix. Rate severity as:
#### Visual Hierarchy
The most important elements should be the most visually prominent — through size, weight, color, and position.
Check for:
Fix pattern: Increase size/weight contrast between hierarchy levels. Primary elements should be 1.5-2x more prominent than secondary. Use the design system's heading scale consistently.
#### Scale & Contrast
Visual weight signals importance. Size differences should be meaningful, not arbitrary.
Check for:
Fix pattern: Use a consistent type scale (e.g., 1.25 or 1.333 ratio). Ensure at least 2px difference between adjacent levels. Icons should match the x-height or cap-height of adjacent text.
#### White Space
Space creates focus, readability, and breathing room. It's not empty — it's functional.
Check for:
Fix pattern: Use the spacing scale from the design system. Section padding should be consistent (e.g., py-16 or py-24). Internal card padding should be at least p-6. Line-height: 1.5-1.7 for body, 1.1-1.3 for headings.
#### Typography Hierarchy
Clear, distinct levels: h1 → h2 → h3 → body → caption. Each should be instantly distinguishable.
Check for:
Fix pattern: Define 5-6 distinct text styles and use them consistently. Each level should differ in at least 2 properties (size + weight, or size + color). Document in the design system.
#### Color Use
Color should be meaningful, restrained, and intentional. Every color should have a job.
Check for:
Fix pattern: Limit accent color to interactive elements and key highlights. Use neutral tones for structure. Each color in the palette should have a clear role (bg, text, accent, muted, border).
#### Proximity — Spacing as Data Structure
Elements close together are perceived as related. Spacing communicates structure.
Check for:
Fix pattern: Inner spacing (within a group) should be notably tighter than outer spacing (between groups). Rule of thumb: inter-group spacing ≥ 2× intra-group spacing.
#### Similarity — Consistent Styling for Same-Type Elements
Elements that look the same are perceived as the same type. Visual consistency = semantic consistency.
Check for:
Fix pattern: Create and enforce component variants: primary, secondary, ghost. Same variant = same visual treatment everywhere. Audit all instances of each component type.
#### Common Region / Uniform Connectedness
Elements within a bounded area (card, panel, background color) are perceived as a group.
Check for:
Fix pattern: Use cards, subtle background shifts, or borders to group related content. One visual container = one logical group. Avoid nesting more than 2 levels deep.
#### Continuity — Alignment & Grid Consistency
The eye follows smooth paths. Alignment creates visual flow and order.
Check for:
Fix pattern: Establish a grid (e.g., 12-column) and align everything to it. Left edges should align vertically across sections. If mixing alignment, do so deliberately (e.g., hero centered, content left-aligned).
#### Closure & Pragnanz — Clean, Minimal, Complete Forms
People perceive the simplest, most complete interpretation. Incomplete or ambiguous shapes create tension.
Check for:
Fix pattern: Every visual element should feel complete and intentional. If something is partially visible (carousel peek), make the clip clearly intentional. Maintain consistent border-radius, shadows, and corner treatments.
#### Fitts's Law — Target Size & Clickable Affordance
Interaction targets should be appropriately sized and visually signal clickability.
Check for:
Fix pattern: All interactive elements need: (1) visual affordance (color, underline, or shape), (2) hover/focus state, (3) adequate size (min 44px touch, 32px click). Primary CTA should be visually larger than secondary.
#### Aesthetic-Usability Effect — Polish = Perceived Usability + Trust
Visually polished interfaces are perceived as easier to use and more trustworthy, even before interaction.
Check for:
Fix pattern: Pick consistent values for border-radius, shadow, transition duration, and apply everywhere. Audit all micro-details. Every pixel matters for trust.
#### Doherty Threshold — Perceived Performance & Sub-Second Feedback
Interactions should feel responsive. Feedback within 400ms keeps users in flow.
Check for:
Fix pattern: Add :active states to all buttons. Use skeleton loaders or fade-ins for async content. Micro-interactions: 150-300ms. Page transitions: 200-400ms. Always show loading state for operations >300ms.
#### Internal Consistency
Same things should look and behave the same way everywhere in the UI.
Check for:
gap-4, another uses gap-6 for same layout)Fix pattern: Extract shared components. Create a component audit spreadsheet: list every instance of buttons, cards, headings, spacing — find and fix deviations. Use design tokens consistently.
Present findings grouped by severity:
## UI Audit Results
### 🔴 Critical — visually broken, damages trust
| Issue | Principle | File | Recommended Fix |
|-------|-----------|------|-----------------|
| ... | ... | ... | ... |
### 🟠 Major — noticeable inconsistency
| Issue | Principle | File | Recommended Fix |
|-------|-----------|------|-----------------|
| ... | ... | ... | ... |
### 🟡 Minor — polish opportunities
| Issue | Principle | File | Recommended Fix |
|-------|-----------|------|-----------------|
| ... | ... | ... | ... |
### ✅ Well done
- [list what the project already does well, referencing which principles it follows]Ask the user: "Should I fix the issues? All issues or only the critical ones?"
Apply fixes directly in the codebase. For each fix:
UI Audit Complete!
✅ [N] issues fixed
⚠️ [N] issues flagged for manual review
Principles applied:
- [list which visual principles were most relevant]
Top 3 highest-impact changes:
1. [change] — [which principle, visual impact]
2. [change] — [which principle, visual impact]
3. [change] — [which principle, visual impact]
Recommended next steps:
- Screenshot comparison (before/after) on key breakpoints
- Cross-browser check (Safari, Firefox, Chrome) for rendering consistency
- Mobile device testing for touch targets and spacing
- Design review with stakeholders on visual hierarchy changes~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.