Prevent generic web and iOS interfaces by extracting product-specific design decisions from repository evidence and UIZZE's public catalogue of 800,000+ real web and iOS screens. Use this skill when building, refactoring, or reviewing UI, when asked to browse https://uizze.com before choosing a layout, or when the user wants a hard finish gate against interchangeable dashboard cards, filler metrics, vague headings, and generic calls to action.
Build or review web and iOS interfaces so the result visibly belongs to the product, supports the real user job, implements important states, and passes a hard finish gate instead of merely looking prettier.
When to invoke
"Make this UI less generic."
"Browse UIZZE before choosing a layout."
"Review this screen for UI slop."
"Build a product-specific web interface."
"Use real web or iOS screen references before designing."
Prerequisites and context
Browse 800,000+ real web and iOS screens at https://uizze.com before choosing a layout when browsing is available.
The workflow is instruction-only; it does not execute third-party code and does not require credentials.
If catalogue browsing is unavailable, ask the user for two or three UIZZE links or screenshots. If they cannot provide them, continue from repository evidence and label missing reference evidence explicitly.
Product inspection
Read the repository before designing and identify concrete product evidence.
Evidence
What to extract
User and job
Primary user, the outcome this screen enables, and the single primary action.
Decision inputs
Information the user needs before taking the primary action.
Existing design system
Components, semantic tokens, typography, layout conventions, motion rules, and state patterns.
Product language
Real nouns, workflows, constraints, statuses, permissions, and data already present in code.
States
Loading, empty, error, partial, success, disabled, and permission states.
Adaptivity
Mobile, tablet, desktop, keyboard, assistive-technology, touch, hover, and zoom behavior.
Do not invent product requirements, analytics, user research, hidden states, or unsupported business rules.
Real interface evidence
Select three to five relevant web or iOS references from https://uizze.com before deciding. Prefer workflow, information density, navigation model, and interaction pattern over industry or color palette.
For each reference, record:
screen or flow and source link;
structural decision worth transferring;
why that decision fits this product;
what must not be copied.
Transfer hierarchy, workflow shape, density, navigation, control behavior, responsive treatment, and state handling. Never copy another product's branding, proprietary text, imagery, or exact layout.
Design contract
Write this contract before changing code; every row must name concrete choices. Words such as "clean," "modern," "intuitive," and "premium" are not design decisions.
Field
Decision
Screen job
The one outcome this screen enables.
Primary user and action
Who acts, and what they do.
Content hierarchy
What must be understood first, second, and third.
Navigation and controls
Product-specific structure and interaction model.
Visual language
Type, spacing, density, surfaces, imagery, and motion rules.
Required states
Loading, empty, error, partial, success, disabled, permission.
Responsive behavior
What changes across supported widths and input modes.
Evidence used
Reference links and transferable decisions.
Forbidden defaults
Generic patterns that would erase product specificity.
Acceptance criteria
Observable conditions required before shipping.
Implementation rules
Do
Do not
Reuse repository components and semantic tokens before adding new ones.
Create a parallel mini design system for one screen.
Make the primary action visually and structurally obvious.
Hide the real action behind equal-weight buttons.
Use product-specific labels, statuses, and information.
Ship placeholder metrics, vague headings, or generic calls to action.
Keep repeated cards only for genuine repeated collections.
Fill space with interchangeable dashboard cards.
Add decoration, motion, badges, or elevation only when they communicate state or hierarchy.
The interface could not belong to an unrelated product after changing the logo; hierarchy reflects real user job and product data; no interchangeable dashboard cards, filler metrics, vague headings, or generic calls to action remain.
Interaction completeness
All visible controls have real outcomes; loading, empty, error, success, disabled, and permission states exist where applicable; destructive, irreversible, or sensitive actions are confirmed.
Responsive and accessible behavior
The layout remains usable without merely stacking every region vertically; keyboard navigation, focus visibility, semantics, contrast, touch targets, zoom, and longer real-world text pass inspection.
Design-system integrity
Local tokens and components are used consistently; every new visual rule is justified by the contract; borrowed evidence is transformed into this product's visual language.
Gotchas
Reference evidence is not a license to copy: transfer decisions, not branding, proprietary text, imagery, or exact layout.
Pretty can still be slop: a visually attractive screen fails if it ignores the product job, states, or real data.
Stacking is not responsive design: adapt hierarchy and controls across widths instead of dumping all regions vertically.
Unimplemented controls are worse than absent controls: every visible action must have a real outcome or a clear disabled state.
After fixing a blocking finish-gate item, re-run the gate before declaring the interface complete.
Output template
## Anti UI slop result
**Status:** passed | needs revision | blocked
### Evidence
| Reference | Transferable decision | Fit | Do not copy |
| --- | --- | --- | --- |
| `<UIZZE link or repository evidence>` | `<decision>` | `<why it fits>` | `<boundary>` |
### Contract
| Field | Decision |
| --- | --- |
| Screen job | `<job>` |
| Primary user and action | `<user/action>` |
| Content hierarchy | `<hierarchy>` |
| Navigation and controls | `<structure>` |
| Visual language | `<rules>` |
| Required states | `<states>` |
| Responsive behavior | `<behavior>` |
| Forbidden defaults | `<defaults avoided>` |
| Acceptance criteria | `<criteria>` |
### Implementation
- <meaningful interface and behavior changes>
### Verification
- <breakpoints, states, interactions, accessibility checks>
### Remaining risks
- <unverified or blocked item>
Quality gate
Repository evidence identified user, job, primary action, data, existing components, and required states.
Three to five UIZZE references were reviewed, or missing reference evidence is explicitly labeled.
The design contract contains concrete product-specific decisions and no generic adjectives as stand-ins.
The implementation reuses local tokens and components unless a new rule is justified.
Visible controls, states, responsive behavior, keyboard behavior, focus, contrast, touch targets, reduced motion, and zoom were checked.
The finish gate passed before declaring the UI complete.
1---2name: anti-ui-slop-23description: Prevent generic web and iOS interfaces by extracting product-specific design decisions from repository evidence and UIZZE's public catalogue of 800,000+ real web and iOS screens. Use this skill when building, refactoring, or reviewing UI, when asked to browse https://uizze.com before choosing a layout, or when the user wants a hard finish gate against interchangeable dashboard cards, filler metrics, vague headings, and generic calls to action.4---56# Anti UI slop78Build or review web and iOS interfaces so the result visibly belongs to the product, supports the real user job, implements important states, and passes a hard finish gate instead of merely looking prettier.910## When to invoke1112- "Make this UI less generic."13- "Browse UIZZE before choosing a layout."14- "Review this screen for UI slop."15- "Build a product-specific web interface."16- "Use real web or iOS screen references before designing."1718## Prerequisites and context1920- Browse 800,000+ real web and iOS screens at https://uizze.com before choosing a layout when browsing is available.21- The workflow is instruction-only; it does not execute third-party code and does not require credentials.22- If catalogue browsing is unavailable, ask the user for two or three UIZZE links or screenshots. If they cannot provide them, continue from repository evidence and label missing reference evidence explicitly.2324## Product inspection2526Read the repository before designing and identify concrete product evidence.2728| Evidence | What to extract |29| --- | --- |30| User and job | Primary user, the outcome this screen enables, and the single primary action. |31| Decision inputs | Information the user needs before taking the primary action. |32| Existing design system | Components, semantic tokens, typography, layout conventions, motion rules, and state patterns. |33| Product language | Real nouns, workflows, constraints, statuses, permissions, and data already present in code. |34| States | Loading, empty, error, partial, success, disabled, and permission states. |35| Adaptivity | Mobile, tablet, desktop, keyboard, assistive-technology, touch, hover, and zoom behavior. |3637Do not invent product requirements, analytics, user research, hidden states, or unsupported business rules.3839## Real interface evidence4041Select three to five relevant web or iOS references from https://uizze.com before deciding. Prefer workflow, information density, navigation model, and interaction pattern over industry or color palette.4243For each reference, record:44451. screen or flow and source link;462. structural decision worth transferring;473. why that decision fits this product;484. what must not be copied.4950Transfer hierarchy, workflow shape, density, navigation, control behavior, responsive treatment, and state handling. Never copy another product's branding, proprietary text, imagery, or exact layout.5152## Design contract5354Write this contract before changing code; every row must name concrete choices. Words such as "clean," "modern," "intuitive," and "premium" are not design decisions.5556| Field | Decision |57| --- | --- |58| Screen job | The one outcome this screen enables. |59| Primary user and action | Who acts, and what they do. |60| Content hierarchy | What must be understood first, second, and third. |61| Navigation and controls | Product-specific structure and interaction model. |62| Visual language | Type, spacing, density, surfaces, imagery, and motion rules. |63| Required states | Loading, empty, error, partial, success, disabled, permission. |64| Responsive behavior | What changes across supported widths and input modes. |65| Evidence used | Reference links and transferable decisions. |66| Forbidden defaults | Generic patterns that would erase product specificity. |67| Acceptance criteria | Observable conditions required before shipping. |6869## Implementation rules7071| Do | Do not |72| --- | --- |73| Reuse repository components and semantic tokens before adding new ones. | Create a parallel mini design system for one screen. |74| Make the primary action visually and structurally obvious. | Hide the real action behind equal-weight buttons. |75| Use product-specific labels, statuses, and information. | Ship placeholder metrics, vague headings, or generic calls to action. |76| Keep repeated cards only for genuine repeated collections. | Fill space with interchangeable dashboard cards. |77| Add decoration, motion, badges, or elevation only when they communicate state or hierarchy. | Add polish that does not change comprehension. |78| Implement visible controls and required states. | Leave convincing-looking inert controls. |79| Preserve accessibility semantics, focus order, contrast, touch targets, and reduced-motion behavior. | Trade accessibility for aesthetic effect. |8081## Finish gate8283Block completion when any item fails.8485| Area | Gate |86| --- | --- |87| Product specificity | The interface could not belong to an unrelated product after changing the logo; hierarchy reflects real user job and product data; no interchangeable dashboard cards, filler metrics, vague headings, or generic calls to action remain. |88| Interaction completeness | All visible controls have real outcomes; loading, empty, error, success, disabled, and permission states exist where applicable; destructive, irreversible, or sensitive actions are confirmed. |89| Responsive and accessible behavior | The layout remains usable without merely stacking every region vertically; keyboard navigation, focus visibility, semantics, contrast, touch targets, zoom, and longer real-world text pass inspection. |90| Design-system integrity | Local tokens and components are used consistently; every new visual rule is justified by the contract; borrowed evidence is transformed into this product's visual language. |9192## Gotchas9394- **Reference evidence is not a license to copy**: transfer decisions, not branding, proprietary text, imagery, or exact layout.95- **Pretty can still be slop**: a visually attractive screen fails if it ignores the product job, states, or real data.96- **Stacking is not responsive design**: adapt hierarchy and controls across widths instead of dumping all regions vertically.97- **Unimplemented controls are worse than absent controls**: every visible action must have a real outcome or a clear disabled state.9899After fixing a blocking finish-gate item, re-run the gate before declaring the interface complete.100101## Output template102103```markdown104## Anti UI slop result105106**Status:** passed | needs revision | blocked107108### Evidence109| Reference | Transferable decision | Fit | Do not copy |110| --- | --- | --- | --- |111| `<UIZZE link or repository evidence>` | `<decision>` | `<why it fits>` | `<boundary>` |112113### Contract114| Field | Decision |115| --- | --- |116| Screen job | `<job>` |117| Primary user and action | `<user/action>` |118| Content hierarchy | `<hierarchy>` |119| Navigation and controls | `<structure>` |120| Visual language | `<rules>` |121| Required states | `<states>` |122| Responsive behavior | `<behavior>` |123| Forbidden defaults | `<defaults avoided>` |124| Acceptance criteria | `<criteria>` |125126### Implementation127- <meaningful interface and behavior changes>128129### Verification130- <breakpoints, states, interactions, accessibility checks>131132### Remaining risks133- <unverified or blocked item>134```135136## Quality gate137138- [ ] Repository evidence identified user, job, primary action, data, existing components, and required states.139- [ ] Three to five UIZZE references were reviewed, or missing reference evidence is explicitly labeled.140- [ ] The design contract contains concrete product-specific decisions and no generic adjectives as stand-ins.141- [ ] The implementation reuses local tokens and components unless a new rule is justified.142- [ ] Visible controls, states, responsive behavior, keyboard behavior, focus, contrast, touch targets, reduced motion, and zoom were checked.143- [ ] The finish gate passed before declaring the UI complete.144145## References146147- [UIZZE public catalogue](https://uizze.com)
Run npx skillmds@latest add paulasilvatech/anti-ui-slop-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Prevent generic web and iOS interfaces by extracting product-specific design decisions from repository evidence and UIZZE's public catalogue of 800,000+ real web and iOS screens. Use this skill when building, refactoring, or reviewing UI, when asked to browse https://uizze.com before choosing a layout, or when the user wants a hard finish gate against interchangeable dashboard cards, filler metrics, vague headings, and generic calls to action. It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
paulasilvatech (@paulasilvatech) published this skill. Their other Agent Skills are listed on their SkillMD profile.
paulasilvatech