Performance Optimization

unity-ecs-patterns is a practical skill for using Unity DOTS, ECS, Jobs, and Burst to design data-oriented gameplay systems, migrate from MonoBehaviour patterns, and optimize CPU-bound simulation with large entity counts.

godot-gdscript-patterns helps Godot 4 users generate and review GDScript with better scene structure, signals, state machines, autoloads, and async loading patterns. Use it to install proven Godot architecture into gameplay systems, UI flows, and maintainable project code.

react-native-architecture helps plan production React Native and Expo apps with routing, state boundaries, native modules, offline-first flows, and Expo vs bare tradeoffs.

react-modernization is a migration-focused skill for upgrading React 16/17 apps to 18+, converting class components to hooks, updating root APIs, and planning staged refactors with codemod-aware guidance.

vercel-react-best-practices is a Vercel Engineering skill that guides AI agents to optimize React and Next.js performance with prioritized rules for waterfalls, bundle size, and rendering.

The optimize skill helps diagnose and improve UI performance across load time, rendering, animations, images, and bundle size. Use it to measure bottlenecks, prioritize fixes, and verify gains for web apps and interactive frontends.

Use the harden skill to make frontend UI more resilient with better error states, empty states, i18n support, RTL handling, overflow fixes, and real-world edge case coverage for production-ready interfaces.

supabase-postgres-best-practices is a Supabase Postgres optimization skill for query tuning, indexing, schema design, RLS performance, locking, and connection management.

next-cache-components teaches Next.js 16 Cache Components: enable cacheComponents, use `use cache`, set cacheLife, apply cacheTag, and stream dynamic sections with Suspense.

The audit-website skill uses the squirrel CLI to audit websites and webapps across 230+ rules for SEO, technical, content, performance, security, links, and site health, then returns actionable LLM-ready reports.