Skip to content
← Tools / Skills
Skill · .md packageEmil Kowalski (community)/10 trustVerified

Emil Kowalski Skills

Design-engineering and animation craft for AI agents — the taste layer that stops agents picking wrong easings and shadows.

Visit official site →

About

Fifteen MIT-licensed skills by designer-engineer Emil Kowalski (ex-Vercel/Linear, author of Sonner) that encode the small motion and design decisions agents routinely get wrong. Covers building animations from scratch (web and React Native/Expo), strict animation review and codebase audits, Apple design principles for the web, mobile-native web fixes, UI-library selection, prototyping, worst-case UI stress testing, and modern Swift. 2.1M total installs on skills.sh.

What it can do

  • animate / animate-expo (build motion from scratch with correct curves, durations, properties)
  • review-animations / improve-animations / find-animation-opportunities (strict review and codebase-wide motion audits)
  • emil-design-eng (general UI polish rules)
  • apple-design, mobile-native, pick-ui-library, prototype, break-ui, ask-sonner, animation-vocabulary, animation-performance, write-swift

Setup

  1. npx skills@latest add emilkowalski/skills (all) or per-skill.
  2. Manual: copy a skills/<name>/ folder into your agent's skills dir.

Prerequisites

Any SKILL.md-capable agent; no accounts or keys needed.

Use cases

designanimationcode

Works with

Any agent (.md)Claude CodeCursorCodex

Trust score — 9/10

/10 trustVerified

Reputable single author with deep design-engineering credibility, MIT license, large adoption (45.6K stars, 2.1M installs), markdown-only with no data access. Docked from top tier because it's a one-maintainer opinionated taste guide rather than a vendor standard — the 'rules' are craft opinions, so results vary with taste fit.

★ 45,670 GitHub stars

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
8.7/10
Maintenance · recency, releases, issue load
7.9/10
Openness · license + docs
10.0/10
Manual review · sanity read + red-flag check
8.0/10
  • · 45,670 GitHub stars (log-scaled vs catalog peers)
  • · last push 1d ago; 0 releases/12mo; 4 open issues
  • · manual review: one-maintainer opinions; some files are placeholders

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.