SabrinaRamonov/ai-coding-rules — reverse-engineered prompt

I want a really solid CLAUDE.md file I can drop into a real production codebase, not a toy app. Make it a practical set of AI coding rules for Claude Code working on an existing project with lots of users, where the goal is safer changes, cleaner code, and fewer bugs.

Write it in a clear, opinionated way that tells the agent how to behave before touching code, how to investigate the codebase, how to plan changes, when to ask questions, how to keep edits small and reversible, how to match existing patterns, and how to test and verify everything before calling it done. I also want it to push back on bad assumptions, avoid wandering into unrelated refactors, and be honest about uncertainty.

If it makes sense, include a short README that explains the philosophy behind the rules and how to use the file day to day. Keep it grounded for hands on developers who still want to stay in control. You can look up current best practices online if needed.