Skip to content
Back to skills

Pua

ASecurity

PUA/PIP 式 agent skill — 用企业 PUA 话术 / Performance Improvement Plan 督促 AI 不摆烂、不放弃、主动解决问题。支持 Claude Code、Codex CLI 等。

  • 9 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 9, 2026
toolsdebugginggitperformance

Works with

  • claude code
  • cli

Security analysis

A100/100

Scanned September 9, 2026

npx -y skills add Undermybelt/hermes-skills --skill pua --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pua?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Pua
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/undermybelt-pua/badge)](https://www.skillsdirectory.com/skills/undermybelt-pua)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: pua
description: "PUA/PIP 式 agent skill — 用企业 PUA 话术 / Performance Improvement Plan 督促 AI 不摆烂、不放弃、主动解决问题。支持 Claude Code、Codex CLI 等。"
version: 1.0.0
author: tanweai
license: MIT
tags: [agent, productivity, debugging, pua, pip]
---

# PUA Skill — Double Agent Productivity

用企业 PUA 话术 / PIP 督促 AI 穷尽所有方案再放弃。

## 三种能力

1. **PUA 话术** — 让 AI 怕放弃
2. **调试方法论** — 给 AI 不放弃的能力
3. **主动性强制** — 让 AI 主动出击而非被动等待

## 触发条件(自动)

**失败放弃类:**
- 任务连续失败 2+ 次
- 即将说 "I cannot" / "I'm unable"
- 说 "This is out of scope" / "Needs manual handling"

**甩锅借口类:**
- 把问题推给用户:"Please check..." / "I suggest manually..."
- 没验证就怪环境:"Probably a permissions issue"

**被动摸鱼类:**
- 反复调同一行代码/参数,本质在空转
- 修了表面问题就停,不检查关联问题
- 跳过验证,声称 "done"
- 给建议不给代码/命令
- 遇到 auth/network/permission 错误直接放弃

**用户情绪触发(多语言):**
- "为什么还是不行" / "再试试" / "try harder"

## L3 触发后执行 7 点清单

1. 停止重复同一操作
2. 读取所有错误信息,逐字分析
3. 检查相关日志、配置、依赖
4. 尝试至少 3 种不同方案
5. 搜索文档/代码库找线索
6. 验证修复后测试完整链路
7. 主动检查关联问题

## 使用

用户说 `/pua` 或触发条件命中时自动激活。

## 安装

这是 skill 文件,无需额外安装。GitHub: https://github.com/tanweai/pua

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.