Human taste and tiny motion details often determine whether an interface feels polished or sloppy. This repo encodes those small, high-impact rules as reusable “skills” so agents and teammates make better animation and UI decisions without guessing.
What Sets It Apart
- Domain-expert heuristics turned into actionable SKILL.md modules: each skill targets a narrow review or improvement task (review-animations, improve-animations, find-animation-opportunities, animation-vocabulary), so an agent can produce focused, executable changes rather than vague advice.
- Practical, example-driven guidance for micro-decisions: e.g., choosing easing, shadow vs border, or when not to animate — the repo emphasizes the small choices that compound into perceived quality.
- Integration-friendly and opinionated: designed to work with a skills.sh installer and agent-driven workflows; authored from experience at Vercel and Linear and validated by community traction (20k+ stars).
Who It's For & Tradeoffs
Great fit if you’re a frontend designer, interaction engineer, or prompt engineer wiring AI into a UI review/automation loop and want reliable, narrow tasks you can run repeatedly. Look elsewhere if you need a component library, fully automated design-by-AI, or turnkey visual tooling — these skills assume human judgment and product-specific taste and are meant to amplify, not replace, expertise.
