Blind A/B testing for AI agent skills, .cursorrules, CLAUDE.md, and system prompts. Detects placebo and harmful instructions.
What it is, and how people use it
Blind A/B testing for AI agent skills, .cursorrules, CLAUDE.md, and system prompts. Detects placebo and harmful instructions. It sits on the Cursor Rules shelf of the catalog, maintained by @sx4im.
Why builders keep it installedCursor Rules benefits
Rules encode your framework choices, folder layout and TypeScript style once. Every new chat starts already knowing how your project is built.
Enforce tests, formatting and typing at the prompt level and the agent's pull requests arrive shaped like your team's, not like a tutorial's.
It is plain text checked into .cursor/rules. No runtime, no account, no server - and the file doubles as documentation for humans.
The same rules files increasingly work in Windsurf, Copilot and Claude Code. A convention you write once pays off on every tool you try next.
Contributorsloading from GitHub…
Languagesestimated from snapshot
- TypeScript 70.0%
- Python 18.0%
- Shell 12.0%