Skip to content

aidrecabrera/bulletproof-react-skill

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 

Repository files navigation

bulletproof-react

skills.sh

An agent skill for bulletproof-react conventions. Feature folders, API layers, state categories, testing, error handling, auth, performance. Knows when to skip the structure too.

I made this because I use bulletproof-react as a baseline for React architecture. It is the most complete set of conventions I have found. But I have also felt the pain of cargo-culting. Applying patterns because they are popular, not because they fit. This skill is designed to avoid that. It checks scale, pushes back when you do not need the full skeleton, and flags when you have overdone it.

Agents should amplify your development workflow and support your judgment, not replace either. (Remember, skill atrophy is real.)

A skill like this is a multiplier for someone who already knows what good React architecture looks like. It augments your development workflow and aids your judgment.

It will not make that call for you. It will remind you of the right questions to ask.

Install

npx skills add aidrecabrera/bulletproof-react-skill

Auto-detects Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and others.

For manual install or other agents, copy skills/ to your agent's skills directory.

Companion tools

  • react-doctor. Scans your codebase for state misuse, performance anti-patterns, architecture violations, and security issues. Run npx react-doctor@latest at your project root.

  • Vercel react-best-practices. 70 performance rules for React and Next.js. Install alongside this skill for performance-specific guidance.

  • Vercel composition-patterns. Component design patterns: compound components, avoiding boolean prop proliferation, React 19 APIs. Install for API design guidance.

Notes

Humans abused useEffect so much that LLMs got trained to overuse them too. Every unnecessary Effect in the training data taught the model this is normal React. The agent is not being lazy. It is repeating what it learned from thousands of examples where an Effect was the wrong answer but nobody flagged them.

GPT-5.6 no longer exhibits that behavior. (source). But even with better models, the reference is still worth having as a reminder. The question is not "can I use an Effect here" but "should I use an Effect here." If there is no external system to synchronize with, calculate during rendering or handle the event directly.

License

MIT. See NOTICE.md.

About

Skill that applies bulletproof-react conventions to your React codebases.

Topics

Resources

Stars

5 stars

Watchers

0 watching

Forks

Contributors

Languages