What happened
Trystan-SA released an MIT-licensed, reverse-engineered system prompt and skill library for Claude Design on GitHub. This open-source project transforms large language models (LLMs) into opinionated, "AI-slop-resistant" design collaborators. The library includes a 20-chapter design philosophy covering content, aesthetics, visual hierarchy, accessibility (WCAG, semantic HTML), interaction, and system thinking, alongside 14 procedural skills for production, extraction, and review, such as accessibility-audit and <a href="/news/2026/6/3/5/ai-outperforms-law-professors" class="text-primary hover:underline">ai</a>-slop-check. It is compatible with various LLMs, including Claude, GPT, Gemini, and local models.
Why it matters
Design teams gain a framework to enforce quality and accessibility standards in AI-generated outputs. The prompt explicitly rejects generic AI design tropes, providing a structured philosophy and auditable skills to guide LLMs toward disciplined, accessible, and consistent visual outputs, reducing reliance on manual correction of "AI slop" and improving compliance with standards like WCAG. Platform engineers can integrate this prompt to standardise AI-assisted design workflows, ensuring adherence to brand guidelines and accessibility requirements. This follows Anthropic's own Claude Design, which launched in April 2026, offering a proprietary visual prototyping tool with code handoff.
