Anti AI Slop UI/UX logo

Anti AI Slop UI/UX

Community
Vanszs
Anti-AI-Slop UI/UX

Overkill negative-prompt skill that stops AI-generated UI from looking like "AI slop" in a single glance. Detects and bans the statistical-mean design defaults LLMs emit (purple gradients, Inter, 3-card grids, glassmorphism, pill CTAs, hype copy) across EVERY component — hero, color, typography, cards, layout, icons, imagery, microcopy, motion, buttons, forms, nav, footer. Use when designing, building, reviewing, or refactoring any UI.

Overview

PublisherVanszs
RepositoryAnti-AI-UI
Skill nameAnti-AI-Slop UI/UX
Stars
20
Forks
6
Bundled files
2
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Vanszs on GitHub. Read the source before you install it.

Installation

Install the Anti AI Slop UI/UX AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Vanszs/Anti-AI-UI/tree/main/Anti-AI-SLop
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Vanszs/Anti-AI-UI.git /tmp/Anti-AI-UI
mkdir -p .claude/skills
cp -r /tmp/Anti-AI-UI/Anti-AI-SLop .claude/skills/vanszs-anti-ai-slop-ui-ux
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Anti AI Slop UI/UX in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Anti AI Slop UI/UX on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Anti AI Slop UI/UX is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Anti-AI-Slop UI/UX — Negative Prompt Skill

Purpose

Produce UI that a discerning designer cannot instantly flag as machine-generated. This skill is a hard filter against "AI slop": the homogenized, brand-agnostic, statistically-average aesthetic that LLMs default to because they predict the most probable token, not the most intentional one.

If you could swap the logo with a competitor and nobody would notice — it's slop.

Root Cause (why this happens)

LLMs generate UI by averaging across training data dominated by Tailwind docs, shadcn/ui examples, abandoned repos with unchanged defaults, and 2020–2024 Dribbble SaaS shots. The output collapses toward the visual mean. This is structural, not a bug (arXiv:2603.13036, "Interrogating Design Homogenization in Web Vibe Coding"). Every decision left unspecified gets filled with the single most frequent pattern. The cure is deliberate, content-specific variation.


⛔ THE INSTANT-DEATH SIGNATURE

Any 3+ of these together = reads as "AI made this" in under 2 seconds. Treat each as a hard block, not a suggestion.

purple/indigo gradient  +  Inter everywhere  +  centered hero + dual CTA
+ 3-card feature grid   +  rounded-2xl on everything  +  gradient pill button
+ "Trusted by 5,000+ teams"  +  "Supercharge your workflow" copy
+ gradient/aurora blobs  +  glassmorphism  +  emoji icons  +  left-border cards

If your output contains this combo, you have generated the canonical AI slop landing page. Stop and regenerate.


How To Use This Skill

When designing/building UI

  1. Read the Per-Component Negative Prompts below for every component you create.
  2. Make a deliberate, stated choice for each of the 5 Slop Axes (color, type, layout, copy, motion) — never accept the default.
  3. Run the Pre-Delivery Slop Audit before handing off.

When reviewing/refactoring UI

  1. Scan against the Slop Severity Table — fix all P0/HIGH first.
  2. Use references/slop-uiux-patterns.md for the exhaustive tell→fix catalog.
  3. Report findings as: tell → why-slop → fix → severity.

The 5 Slop Axes (decide each one explicitly)

AxisSlop default (BANNED)Required action
Colorindigo-500/violet/purple accent, purple→pink/blue→cyan gradientPick ONE brand color (NOT purple/indigo). Define 60-30-10 palette.
TypeInter/Geist for everything, gradient headline, font-extrabold tracking-tighterDistinctive display font ≠ Inter. Solid headline. 2–3 weights, modular scale.
LayoutCentered everything, max-w-7xl, uniform py-20, symmetric 3-col gridVary container widths, section padding, grid structure. ≥2 asymmetric moments.
Copy"Supercharge/Unlock/Seamless", "Get Started", "Trusted by X+ teams"Concrete verbs + real numbers + specific CTAs. No buzzwords, no fake proof.
Motionhover:scale-105, universal fade-in-up, 500ms, no reduced-motion≤200ms, transform/opacity only, prefers-reduced-motion mandatory.

PER-COMPONENT NEGATIVE PROMPTS

🦸 HERO — never generate

  • ❌ Centered headline + subhead + symmetric dual buttons (filled primary + ghost secondary)
  • ❌ Gradient bg-clip-text headline (purple→pink/blue)
  • ❌ Eyebrow pill badge above H1 (✨ Now with AI, 🚀 New)
  • ❌ Aurora/gradient blobs (blur-3xl purple/blue circles) behind content
  • ❌ Perspective-tilted dashboard mockup with glow (rotateY(-5deg) shadow-2xl)
  • ❌ Greyed fake "Trusted by 5,000+ teams" logo strip
  • ❌ h-screen hero with only text + button (PowerPoint-slide emptiness)
  • ✅ Do instead: left-aligned headline, asymmetric split (text left / real product screenshot right), one dominant CTA + a text-link secondary, solid background, specific value-prop copy.

🎨 COLOR & GRADIENT — never generate

  • ❌ indigo-500/violet-500/purple-500/#6366f1/#8b5cf6/#a855f7 as primary accent
  • ❌ from-purple-500 to-pink-500, from-indigo-500 to-purple-600, from-blue-500 to-cyan-500
  • ❌ Gradient on text AND button AND border AND background simultaneously
  • ❌ Neon glow box-shadow: 0 0 40px rgba(purple) on cards/buttons/text
  • ❌ Slate-900 + indigo-accent "default dark mode"
  • ❌ Low-contrast text-gray-400 body on white/dark (fails WCAG 4.5:1)
  • ✅ Do instead: one intentional brand hue, 60-30-10 distribution, flat solid surfaces, functional shadows (shadow-sm/md, neutral), body text ≥ gray-700 (light) / gray-200 (dark).

🔤 TYPOGRAPHY — never generate

  • ❌ Inter/Geist for both display and body with no second face
  • ❌ Gradient bg-clip-text text-transparent headlines
  • ❌ text-5xl font-extrabold tracking-tighter formula on every heading
  • ❌ 5–6 font weights; arbitrary/flat type scale; centered multi-line body text
  • ❌ tracking-widest uppercase eyebrow above every section
  • ❌ Playfair Display + Inter "tasteslop" pairing; monospace for non-code "tech vibes"
  • ✅ Do instead: distinctive display font (Fraunces, Bricolage Grotesque, Satoshi, Cabinet Grotesk), modular scale (1.25), max 2–3 weights (400/500/700), leading-tight headings / leading-relaxed body, max-w-prose (65ch), left-aligned paragraphs.

🃏 CARDS — never generate

  • ❌ Identical 3-column feature grid (icon-in-rounded-square + title + blurb)
  • ❌ Uniform rounded-2xl shadow-lg border on every card
  • ❌ hover:scale-105 (shifts layout) or dramatic -translate-y-2 lifts
  • ❌ Glassmorphism cards (bg-white/10 backdrop-blur-lg)
  • ❌ Gradient-border cards; border-l-4 border-purple-500 left accent
  • ❌ Emoji icons; identical pricing cards with "Most Popular" ring; fake-avatar testimonials
  • ✅ Do instead: vary card size/weight (one larger, one image-led), elevation hierarchy, hover:shadow-md/hover:border-primary color transitions only, solid surfaces, real logos over fake faces, max 5 pricing features.

📐 LAYOUT & SPACING — never generate

  • ❌ Everything centered in max-w-7xl mx-auto
  • ❌ Uniform py-20 section rhythm (no spatial hierarchy)
  • ❌ Symmetric grid-cols-3 for every section
  • ❌ Endless alternating image-left/text-right
  • ❌ Identical section anatomy repeated (heading → subtext → 3-col → CTA ×7)
  • ❌ One gap-6 token for nav, cards, forms, and sections alike
  • ✅ Do instead: vary container widths by content (max-w-2xl text / max-w-6xl grids / full-bleed breaks), vary section padding, mix grid structures (2/4/asymmetric spans), one dominant element per section, contextual gap scale (4px inline → 16px cards → 48px sections), ≥2 deliberate asymmetric moments per page.

🎯 ICONS & IMAGERY — never generate

  • ❌ Emoji as functional icons (🚀⚡✨🎨💡)
  • ❌ Icons in gradient rounded squares; rainbow per-icon coloring
  • ❌ Mixed icon sets/variants/stroke widths
  • ❌ Corporate-Memphis / Alegria blob-people illustrations (unDraw/Humaaans)
  • ❌ AI-generated hero images with artifacts; generic stock "diverse-team-at-laptop"
  • ❌ Glow behind icons; oversized (>48px) decorative line icons; status dots that map to nothing
  • ✅ Do instead: ONE icon library + ONE variant, monochrome/2-color max, sized 16–24px UI, real product screenshots over illustration, every status dot maps to a documented state.

✍️ MICROCOPY & CONTENT — never generate

  • ❌ Hype verbs: "Supercharge, Unlock, Revolutionize, Elevate, Seamless, Effortless, in seconds"
  • ❌ Filler: "ensuring", "leveraging", "robust/powerful/scalable", "cutting-edge", "for modern teams"
  • ❌ Generic CTAs: "Get Started", "Learn More" everywhere
  • ❌ Fake testimonials ("Sarah M., CEO"), invented metrics, "Trusted by X,000+"
  • ❌ Em-dash overuse, uniform paragraph length, "In today's landscape…" openers, emoji bullets, leftover Lorem ipsum
  • ✅ Do instead: say what it does for whom with concrete numbers, specific CTAs ("Import your data"), real or clearly-placeholdered proof, varied sentence rhythm, a human voice with edge.

🎬 MOTION — never generate

  • ❌ Universal fade-in-up on scroll (AOS on everything)
  • ❌ hover:scale-105 on every card; gratuitous Framer-Motion wrappers
  • ❌ Infinite floating/bobbing blobs; marquee logo scrollers; animated gradient backgrounds
  • ❌ Spring bounce on everything; typewriter effects; 500ms+ transitions
  • ❌ NO prefers-reduced-motion support (this is an a11y FAILURE, not just slop)
  • ✅ Do instead: animate ≤30% of elements, 100–150ms hovers / 200–300ms entrances / faster exits, transform+opacity only, ease-out in / ease-in out, mandatory prefers-reduced-motion: reduce, max 0–1 ambient infinite animation (must pause on hover).

🔘 BUTTONS / FORMS / NAV / FOOTER — never generate

  • ❌ Gradient primary + ghost secondary pair; emoji-in-button; rounded-full pills everywhere; oversized px-8 py-4 text-lg
  • ❌ Forms with no validation/error/success states; heavy rounded-xl shadow glass inputs; toggle overuse; generic gradient "newsletter" band
  • ❌ Floating glassmorphic backdrop-blur navbar; shadcn icon+label sidebar clone for non-dashboards; premature sticky CTA bar
  • ❌ 4-column mega-footer of dead # links + social icons you don't use; dark-pattern cookie banner
  • ✅ Do instead: one solid-color primary per section + text-link secondary, rounded-md/lg, px-4 py-2 text-sm; inputs border border-gray-300 rounded-md with real error/success states; fixed top-0 border-b solid nav; minimal footer with only real links; equal-weight cookie choices (if needed at all).

Style Guardrails (positive defaults)

  • 8px spacing system; internal spacing ≤ external spacing (proximity = grouping).
  • One dominant focal element per section; 60-30-10 color distribution.
  • One primary action per section. Restrained emphasis.
  • Radius scale, not one value: inputs/badges rounded-md, cards rounded-lg, modals rounded-xl, rounded-full only when semantically a pill/avatar.
  • Real product UI, real data, real screenshots — never demo/placeholder UI.
  • WCAG AA contrast minimum (4.5:1 body, 3:1 large/UI). Test light AND dark.

Slop Severity Table

SeverityMeaningExamples
P0 (auto-block)Single-handedly screams AI; never shippurple/indigo gradient, gradient-text headline, emoji icons, left-border cards, invented metrics, no reduced-motion
HIGHStrong tell; fix before deliveryInter-everywhere, centered dual-CTA hero, 3-card grid, rounded-2xl uniformity, pill gradient CTA, fake "Trusted by", glassmorphic nav, no form validation states, dead-link mega-footer, hype copy, hover:scale-105
MEDIUMTrained eyes noticeperspective mockup+glow, bento misuse, uniform py-20, em-dash overuse, toggle overuse, oversized buttons, illustration-pack sameness
LOWSubtle but cumulative0.1-opacity shadows, Space Grotesk fallback, wide letter-spacing, count-up numbers

Pre-Delivery Slop Audit (run every time)

Color & Type

  • No indigo/violet/purple accent unless it is the real brand
  • No gradient text; ≤1 gradient element per section
  • Display font is NOT Inter/Geist; ≤3 font weights; modular scale
  • Body text passes 4.5:1 contrast (light + dark)

Layout & Cards

  • Container widths vary by content; section padding varies
  • No identical 3-col grid repeated; ≥1 dominant element per section
  • Radius is a differentiated scale, not one uniform value
  • No border-l-4 accent; no uniform glassmorphism; no emoji icons

Copy

  • No "Supercharge/Unlock/Seamless/for modern teams"; CTAs are specific
  • Every metric is real or clearly marked placeholder; no fake testimonials
  • No leftover Lorem ipsum / [placeholder] / TODO

Motion & A11y

  • prefers-reduced-motion: reduce handled everywhere
  • Transitions ≤200ms, transform/opacity only, no hover:scale-105 in grids
  • Forms have label + error + success states; focus states visible
  • Images have alt text; color is not the only indicator

Gestalt check

  • Could I swap the logo with a competitor and nobody notices? → if yes, FAIL
  • Does the Instant-Death Signature appear (3+ tells)? → if yes, regenerate
  • Would this pass a senior designer in a 2-second glance? → if no, fix

Output Expectations

The final UI must read as: intentional, branded, content-aware, accessible, production-ready — and unmistakably NOT machine-averaged. Every spacing value, color, font, grid, and animation must be defensible with a content-specific reason. If you cannot explain why a choice was made, it is slop — remove or replace it.

For the exhaustive tell → why → fix → severity catalog across all components, see references/slop-uiux-patterns.md.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Anti AI Slop UI/UX AI skill do?

Overkill negative-prompt skill that stops AI-generated UI from looking like "AI slop" in a single glance. Detects and bans the statistical-mean design defaults LLMs emit (purple gradients, Inter, 3-card grids, glassmorphism, pill CTAs, hype copy) across EVERY component — hero, color, typography, cards, layout, icons, imagery, microcopy, motion, buttons, forms, nav, footer. Use when designing, building, reviewing, or refactoring any UI.

Why use Anti AI Slop UI/UX on TypingMind?

Because you install it once and use it with any model. Anti AI Slop UI/UX is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Anti AI Slop UI/UX in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Vanszs/Anti-AI-UI/tree/main/Anti-AI-SLop. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Anti AI Slop UI/UX?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Anti AI Slop UI/UX?

As many as you like. As long as a model supports skills, you can use Anti AI Slop UI/UX with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Anti AI Slop UI/UX AI skill free?

Yes. It is published on GitHub by Vanszs under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇