Pua En

This is performance review rhetoric weaponized into a debugging protocol. It triggers when your AI fails twice, starts suggesting you do manual work, or goes passive instead of using its tools to investigate. The skill enforces three rules: exhaust all options before giving up, investigate with available tools before asking questions, and proactively check related issues instead of doing the bare minimum. Each failure escalates the pressure from "verbal warning" to formal PIP language borrowed from Amazon and Google performance culture. It's absolutely unhinged and surprisingly effective if you're stuck in a loop where Claude keeps半吊子-ing solutions or saying "I cannot" without actually trying. Works for any task type: code, configs, deployment, research, whatever.

Install to Claude Code

npx -y skills add tanweai/pua --skill pua-en --agent claude-code