emilkowalski/skills

Selection

Select when

• You want reusable context for building polished web or Expo interfaces.
• You need strict animation review criteria with concrete easing, duration, and accessibility rules.
• You want a read-only motion audit that yields self-contained implementation plans.
• You need several genuinely different UI prototypes for live comparison.
• You want Apple-inspired motion and interface principles translated to web technologies.
• You need current Swift guidance spanning Swift 6 concurrency, performance, and testing.
• You use Sonner and need setup, styling, or troubleshooting guidance.
• You want an opinionated frontend-library recommendation rather than a long option list.

Boundaries

Avoid when

• You need a standalone design or animation application.
• You need the repository itself to execute browser, device, or SaaS actions.
• You need general-purpose code review unrelated to motion.
• You want an animation audit that directly edits production source code.
• You need neutral or multi-option library comparisons rather than curated recommendations.
• You cannot validate motion on real devices or through visual inspection.
• You need guaranteed compatibility with a specific Claude Code version, because none is stated in the supplied files.
• You require verified claims for every external API version referenced by the skills.

Strengths

Capabilities

Risk profile

Risks and limitations

• Uncertain: only selected first-party files were supplied, so uninspected skill dependencies and supporting references may add constraints.
• Uncertain: the README documents installation through the skills CLI but does not explicitly name Claude Code or a minimum compatible version.
• Guidance is intentionally opinionated and may conflict with an existing product motion language or dependency policy.
• Some recommendations are version-sensitive, including Swift, Expo, Reanimated, and browser-platform guidance.
• Several workflows depend on visual judgment that cannot be verified from source code alone.
• The prototype workflow assumes an existing dev server or a browser-openable static HTML context.
• The audit workflow is deliberately read-only and requires a separate executor for planned changes.
• The library picker covers only its curated list and may not fit projects standardized on alternatives.
• External links and referenced APIs may change independently of this repository.
• No automated compatibility, security, or correctness test evidence appears in the supplied files.

The supplied repository content is Markdown guidance rather than an evidenced executable service. Some skills can direct a host agent to write code, inspect a project, or install dependencies when the user requests implementation, so normal code review, dependency review, least-privilege execution, and host confirmation controls remain necessary. No sandbox, permission model, or security audit is documented in the supplied files.