Ezra144israel/governed-agent-skills ? reverse-engineered prompt

Build me a governance package for AI coding agents that I can drop into tools like Claude Code, Codex, Cursor, and similar apps.

I want it to give the agent a clear way of working, with a constitution, a reasoning method, planning discipline, test and verification standards, and a final approval step so the agent that writes code is not the one that approves it. Please also include a small enforcement hook that blocks obviously destructive shell commands before they run, plus a simple way to install and use everything from the repo.

Make it feel polished and ready to publish, with clear docs for manual install and optional progressive loading, and a simple live overview page or demo if that fits. Keep the scope honest, focused on agent workflow governance, and make sure the skills can be installed one by one, with sensible dependencies between them. If you need to check current docs for the agent platforms, go look them up online first.