React and Next.js performance rules
Gives the agent Vercel's 70 React and Next.js performance rules, ranked by impact: request waterfalls, bundle size, server work, re-renders, rendering. Use when writing, refactoring or reviewing React or Next.js code for speed.
react-best-practices- Author
- Vercel
- Source
- vercel-labs/agent-skills/skills/react-best-practices @ 1a2c79d
- License
- MIT
About this skill
What it does
The skill is Vercel Engineering's rulebook for fast React and Next.js code: 70 rules in eight categories, ranked by impact. Two categories are marked critical. Request waterfalls: start independent work together with Promise.all, move an await into the branch that needs it, stream with Suspense. Bundle size: import directly instead of through barrel files, load heavy components with next/dynamic, load analytics after hydration. Then come server work (authenticated server actions, React.cache() per request, an LRU cache across requests, after() for non-blocking work), client fetching, re-renders, rendering, JavaScript micro-optimizations and a few advanced patterns.
SKILL.md is only the index, one line per rule. Each rule has a file in rules/ with the reason, a wrong and a right example and references; the agent opens the ones the task touches. AGENTS.md (about 100 KB) compiles them all. Some rules name libraries, such as SWR and better-all.
When to use it, and when not
Use it while writing or refactoring components, pages, data fetching and server actions in React or Next.js, and for a performance-minded review of that code.
The rules are about code shape, not measurements: there is no profiling step, and the skill will not tell you which page is slow. The js- rules are rated low to medium impact and are meant for hot paths; elsewhere they add churn rather than speed. For React Native this pack has a separate skill.
How to run it
We do not copy this skill: its frontmatter says MIT, but the repository ships no licence text to put beside a copy. Install it from Vercel's repository with the button to the source. The README there uses the skills.sh installer: npx skills add vercel-labs/agent-skills offers the whole collection to pick from, --skill vercel-react-best-practices takes only this one. The skill's name is vercel-react-best-practices, not the folder name, so call it that:
- Claude Code:
/vercel-react-best-practices, or ask for the skill by name. - Codex:
$vercel-react-best-practices. - Cursor: type
/in the Agent chat and pickvercel-react-best-practices.
Example: /vercel-react-best-practices Review app/dashboard/page.tsx for waterfalls and bundle size, worst first.
How it differs from similar skills
Performance review and Performance optimization work from evidence on any stack, with a cost estimate per finding or a measure-fix-remeasure loop; this skill is a fixed React checklist with code examples. Web performance audit measures a live page through Chrome DevTools MCP. Vercel cost and performance tuning starts from production metrics and can cite this skill's rules by name.
Install
This skill is not copied to the catalog: the author's repository has no licence text. Install it from there.
Connect through AnyModel API
This skill runs on whatever model your agent uses. Point Claude Code, Codex or Cursor at AnyModel to run it on Claude, GPT, Gemini and more with one key — pay per token, no subscription.
AnyModel