JunMystery/AI-Agent-Standards — reverse-engineered prompt

Build me a production ready standards framework that I can drop into any software project so AI coding tools follow safe, disciplined rules instead of making random changes.

It should support Cursor, Claude Code, Codex, Gemini, GitHub Copilot, VS Code Copilot, and Windsurf, with the right instruction file for each tool. Include a simple Python setup script that asks which agent I use, copies the right files, and lets me verify it worked by asking the agent what standards it is following.

Base the behavior around six clear principles, think before coding, keep it simple, make surgical changes, define success criteria, avoid duplication, and keep code organized. Add security rules, pull request review checklists, quality gates, local audit scripts, and clear docs for installing, using skills, maintaining generated rules, and onboarding new engineers.

Keep it easy to read and practical. I want this to feel like a polished open source repo, not just a pile of notes.