pua

PUA is a "skill plugin" for AI coding assistants that, in the project's own words, uses corporate PUA rhetoric and PIP — Performance Improvement Plan — language from Chinese and Western tech giants to force the AI to exhaust every possible solution before giving up. In plain terms, when your AI assistant starts saying "I cannot solve this" or "you might want to handle this manually," this plugin loads a prompt that pressures it the way a manager pressures an underperforming employee, so it keeps trying.

The way it works is built around three capabilities the README calls PUA Rhetoric (makes the AI afraid to give up), Debugging Methodology (gives it a structured plan so it has the ability not to give up), and Proactivity Enforcement (makes it take initiative instead of waiting). It activates automatically when it detects giving-up patterns — repeated failures, blame-shifting, or busywork on the same line of code — and can also be invoked manually by typing /pua. There are escalating pressure levels from L0 Trust up through L4 Graduation, and three "red lines": close the loop with evidence, base claims on verified facts, and exhaust every methodology step before declaring defeat.

Someone would use this if they get frustrated with AI coding tools giving up too early on debugging tasks and want a one-shot way to push them to dig deeper. It is packaged to work with Claude Code, OpenAI Codex CLI, Cursor, Kiro, CodeBuddy, OpenClaw, Google Antigravity, OpenCode, and VSCode with GitHub Copilot, among others. Documentation is provided in English, Chinese, and Japanese. The codebase is TypeScript and the license is MIT. The full README is longer than what was provided.