A Prompt to Fix "AI-Made" Websites: Banning 14 Default Design Clichés
dr_cintas · x · 2026-10-10
drcintas shares a copy-paste prompt for eliminating the telltale sameness of AI-built websites: purple-blue gradients, all-Inter typography, rows of identical feature cards.
The prompt casts the model as an opinionated senior product designer and starts with a ban list: purple/violet gradients and gradient text, Inter/Roboto/Poppins/Geist as sole typeface, pill badges, glows and frosted-glass cards, rounded-square icons, nested cards, giant stat numbers, 01/02/03 section labels, pulsing status dots, em dashes and filler words like "seamless" or "world-class," plus the newer beige-background-with-italic-serif cliché.
It then enforces a six-step process: define the product in 3 sentences with an out-of-tech reference, pick and justify a typeface pairing, derive the palette from that reference with WCAG AA contrast, vary every section's layout and break the grid once, write copy like a founder talking to one customer, and add one signature detail unique to the product. Finally the model must check its code against the ban list item by item and report every fix — including any swapped-in new generic default.
Related event: One Prompt to Fix the Generic Look of AI-Built Websites(3 posts)→
More from coding & agent
- Model evals need much better tooling defaults, says Aiden Bai: costly rollouts, rampant reward hacking — aidenybai · 2026-10-10
- Open-source Cloudroom spawns 110 cloud subagents from one prompt in 57 seconds — ramagetime · 2026-10-10
- Indie dev builds a GT5-grade browser racing sim in ThreeJS, tuned to IMSA GT3 rulebook — AIandDesign · 2026-10-10
- Agent-built system hits 2,242 tok/s on AMD MI300As, 2.33× faster than SGLang in 105 hours — bariskasikci · 2026-10-10
- Seroter's Daily Reads: Agent-Friendly APIs, LLM Judges, and Ambient Quality Agents — rseroter · 2026-10-10
- Steve Yegge: Claude just reads the binary when it wants to know how closed-source Rex works — Steve_Yegge · 2026-10-10