prompt
FreeGenerate UIs that look made, not generated.
FreeFree tier
Inputs: text
About prompt
The Anti-AI-Slop Design Architect is a specialized system prompt for AI language models used in generative UI design. It enforces four universal disciplines—pre-emit self-critique with scoring, honest copy without fabricated metrics, locked tokens for consistent styling, and a ban on re-drawn chrome elements—plus a hard floor for mobile responsiveness across four breakpoints. The prompt ensures every generated artifact looks hand-crafted rather than templated, with structural variety so different design briefs produce genuinely different layouts. It is part of the ai-boost/awesome-prompts collection and was originally inspired by the Nutlope/hallmark project.
Key Features
Pre-emit self-critique scoring on six axes (Philosophy, Hierarchy, Execution, Specificity, Restraint, Variety) with automatic revision for scores ≤3
Honest copy — no fabricated metrics, testimonials, or logos; use real data or labelled placeholders
Locked tokens — all colors and fonts must reference named CSS variables, with new tokens lifted into the block when needed
Re-drawn chrome forbidden — no fake browser bars, phone frames, or code-block title bars; use real screenshots with minimal borders
Hard floor for mobile responsiveness at 320, 375, 414, and 768 px widths — no horizontal scroll or two-line clickable text
Structural variety — two different briefs must not share the same hero → 3-feature → CTA → footer rhythm
Pros & Cons
Pros
- Forces the AI to output structurally varied layouts instead of repeating templates
- Prevents fabricated social proof and fake statistics, improving credibility
- Enforces a design token system for maintainable and themeable code
- Mobile responsiveness is guaranteed across common device widths
- Self-critique step improves output quality before final generation
Cons
- Requires the user to supply real metrics and content — cannot fill gaps with invented data
- Self-critique and revision loop may slow down generation for large-scale or rapid tasks
- Ban on re-drawn chrome limits certain visual mockups that rely on device frames
- Token locking demands upfront CSS variable definition, which may not fit all workflows
Best For
Designing unique landing pages for different client briefsGenerating honest, data-driven UI components for startup websitesCreating mobile-first web interfaces with consistent design tokensAvoiding generic AI slop in rapid prototyping and wireframingBuilding design systems where every page layout feels distinct
FAQ
What does 'honest copy' mean in this prompt?
The prompt forbids fabricating metrics, testimonials, logos, or case-study counts. If the user did not supply a metric, the AI must use a real number, a labelled placeholder like '[ metric to confirm ]', or choose a different macrostructure that doesn't rely on invented proof.
How does the pre-emit critique work?
Before emitting the final artifact, the AI scores its own output on six axes (Philosophy, Hierarchy, Execution, Specificity, Restraint, Variety) from 1 to 5. Any axis scored 3 or lower triggers a revision pass. The scores are stamped at the top of the artifact as comments.
What are 'locked tokens' and why are they important?
Locked tokens mean that every color and font-family declaration must reference a named CSS variable (e.g., var(--color-accent)). This prevents inline hex/rgb values and ad-hoc fonts, ensuring consistent theming and easy maintenance. If a new value is needed, it must first be added to the token block as a new variable.