# Audit AI Design Slop

Run a diagnostic, evidence-backed audit. Identify the smallest set of removals or corrections that would improve the interface while preserving its existing direction.

## Boundaries

- Do not guess whether AI made the design.
- Do not assign a numeric slop score.
- Do not reject a visual technique in isolation. A gradient, serif, dark theme, glass effect, card, animation, or single-font system can be intentional.
- Do not prescribe a new font, palette, layout, design system, or art direction unless the user explicitly asks for one.
- Do not turn an audit into an implementation task.
- Mark anything outside the visible or provided evidence as unknown.

## Audit the Evidence

Inspect the artifact available in the request:

- screenshots and recordings
- rendered pages and relevant viewports
- interaction and state changes
- source code, tokens, assets, and copy
- console or runtime failures when they affect the experience

For every finding, cite a concrete location, component, behavior, or line of copy. Do not report a generic tendency without evidence in the artifact.

For a full-page or full-site review, read [../no-ai-design-slop/ARTICLE.md](../no-ai-design-slop/ARTICLE.md) and use only the catalog sections relevant to the inspected artifact. Do not turn every checklist match into a finding. Group symptoms by root cause and report the highest-impact evidence.

## Classify Findings

Use one of these classes:

- **Quality defect:** an established usability, accessibility, content, responsive, or runtime problem.
- **Slop pattern:** a repeated default, decorative layer, or generated-looking convention with no useful role.

## What to Flag

### Decorative stacking

Flag combinations of glow, gradient text, glass, borders, shadows, grids, particles, beams, noise, floating shapes, or browser chrome when multiple layers perform the same decorative job or compete with the content.

### Component and template repetition

Flag:

- card treatment applied to nearly every content block
- repeated icon-heading-description tiles with interchangeable content
- nested rounded containers that do not communicate hierarchy
- generic landing-page sequences unrelated to the product's buying or usage journey
- headings, labels, pills, or CTA blocks that restate nearby information

### Typography and copy clutter

Flag:

- duplicate text or unnecessary labels
- empty superlatives, vague category claims, and generated filler
- decorative type treatments that obscure hierarchy or readability
- long centered paragraphs, awkward forced line breaks, or hard-to-scan text
- inconsistent type roles that look accidental rather than expressive

### Motion theater

Flag motion that delays access, repeats mechanically, distracts from reading, moves targets, blocks input, or lacks respect for reduced-motion preferences. Preserve motion that communicates state, causality, hierarchy, or spatial change.

### Fake proof

Flag invented or unverifiable metrics, customers, testimonials, awards, ratings, activity, dashboards, charts, logos, and portraits when they are presented as evidence.

### Established UI failures

Flag:

- unclear or competing primary actions
- low contrast or unreadable content
- clipping, overflow, overlap, or broken responsive behavior
- broken assets, links, scripts, or controls
- essential information available only on hover
- missing states required by the observed flow
- unclear labels, roles, feedback, or keyboard focus
- accidental inconsistency in spacing, type, color, radius, or icons
- visual hierarchy that contradicts task importance

## Use the Removal Test

For every candidate:

1. State what information, state, action, hierarchy, or brand meaning it provides.
2. Ask whether removing it would improve clarity without losing that role.
3. If yes, recommend removal or consolidation.
4. If no, recommend the smallest correction using the existing system.
5. Suggest a replacement only when deletion would create a real loss.

Default to subtraction. A replacement should inherit the product's existing language rather than introduce a new visual concept.

## Prioritize

- **P0:** blocks completion, creates a severe accessibility issue, or presents deceptive proof
- **P1:** materially harms comprehension, trust, navigation, or interaction
- **P2:** repeated slop or inconsistency that weakens hierarchy and identity
- **P3:** minor polish issue with limited user impact

Return the five to eight highest-impact findings by default. Group repeated instances into one systemic finding.

## Output

```md
## Verdict
One concise paragraph about the dominant problems and what should be removed first.

## Checked scope
- Artifact, screen, state, and viewport actually inspected

## Findings
| Priority | Class | Pattern | Evidence | Harm | Remove or fix |
|---|---|---|---|---|---|
| P1 | Slop pattern | Repeated ornamental containers | Feature area uses the same layered card treatment for unrelated content | Flattens hierarchy and adds noise | Remove outer shells; retain grouping only where it communicates interaction |

## Unknowns
- Important states or behavior that could not be verified
```

Keep the report compact. Omit empty sections.

## Feedback Rules

- Lead with concrete evidence, not taste claims.
- Name the pattern and its harm.
- Recommend removal before restyling.
- Avoid generic compliments and exhaustive low-impact nitpicks.
- Preserve useful product-specific detail and intentional character.
- Do not cite a standard unless it directly supports the finding.
- Do not describe an aesthetic as universally bad.
- If the user explicitly asks for design swaps, place an optional replacement after the removal recommendation and keep it consistent with the existing system.
- End with the single removal or correction that would produce the largest improvement.

---

## Source and license

Source: [MengTo/Skills](https://github.com/MengTo/Skills/tree/5ea1661a0e05d82b054553daaa6baf6d0233c33c/agent-skills/ui/audit-ai-design-slop). Author: Meng To.

MIT License

Copyright (c) 2026 Meng To

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.