Taste Skill
The anti-slop frontend framework for AI agents — tunable design dials and style skills that kill generic AI-looking UIs.
About
A portable skill suite that upgrades AI-built interfaces with stronger layout, typography, motion, and spacing instead of boilerplate output. The default design-taste-frontend skill (currently v2 experimental, with v1 pinnable) reads the brief, infers a design language, and tunes VARIANCE/MOTION/DENSITY dials; sibling skills cover redesign audits, high-end/minimalist/brutalist styles, and image-generation reference boards for web, mobile, and brand kits.
What it can do
- design-taste-frontend (brief inference, design-system map, pre-flight check, GSAP skeletons)
- redesign-existing-projects (audit-before-redesign)
- gpt-taste (stricter GPT/Codex variant)
- high-end-visual-design, minimalist-ui, industrial-brutalist-ui, stitch-design-taste
- image-generation skills (imagegen-frontend-web, imagegen-frontend-mobile, brandkit) producing reference comps only
Setup
- All skills:
npx skills add https://github.com/Leonxlnx/taste-skill - Single skill:
npx skills add https://github.com/Leonxlnx/taste-skill --skill "design-taste-frontend" - Claude Code plugin:
/plugin marketplace add Leonxlnx/taste-skill→/plugin install taste-skill@taste-skill
Prerequisites
Any SKILL.md-capable agent; image-gen skills pair with ChatGPT Images or similar generators.
Use cases
Works with
Trust score — 9/10
Very widely adopted (95K stars), MIT, active team with changelog, framework-agnostic, markdown-only. Minor caveats: the default skill is labeled v2 experimental (v1 is pinnable); the README carries sponsor placements, signaling a commercial ecosystem around the project — the skills themselves are free and MIT.
★ 95,074 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 9.2/10
- Maintenance · recency, releases, issue load
- 7.9/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · 95,074 GitHub stars (log-scaled vs catalog peers)
- · last push 1d ago; 0 releases/12mo; 6 open issues
- · manual review: v2 experimental rewrite; sponsor placements in docs
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.