Skip to content
Back to search

pua-skills

MarketplaceWorks with Claude CodeGitHub

PUA skills: opt-in productivity coaching for try-harder moments, repeated-failure troubleshooting, evidence-first completion checks, and Ding-style workplace reminders.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/tanweai-pua/badge.svg)](/listings/tanweai-pua)

HTML

<a href="/listings/tanweai-pua"><img src="/api/listings/tanweai-pua/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Vercel Skills CLI

Vercel Skills CLI is a general installation method for skills and is not tied to a specific AI tool. This English README installs the English skill:

npx skills add tanweai/pua --skill pua-en

If the current session does not pick up the new skill immediately, restart your AI tool.

Claude Code

claude plugin marketplace add tanweai/pua
claude plugin install pua@pua-skills

**To update:**

# Refresh marketplace cache first, then update (skipping the first step may install an old cached version)
claude plugin marketplace update
claude plugin update pua@pua-skills

**Developer install (source):**

git clone https://github.com/tanweai/pua ~/.claude/plugins/pua

Then manually register in `~/.claude/plugins/installed_plugins.json`:

{
  "version": 2,
  "plugins": {
    "pua@pua-skills": [
      {
        "scope": "user",
        "installPath": "/Users/<you>/.claude/plugins/pua",
        "version": "<installed-version>"
      }
    ]
  }
}

Replace `<installed-version>` with the version in the cloned `plugin.json`; do not copy a stale cache version.

> **Windows:** use `C:/Users/<you>/.claude/plugins/pua` as `installPath`.

Restart Claude Code. To update: `git pull` inside `~/.claude/plugins/pua`.

**Optional: bare command alias (requires plugin installed above — adds `/pua` without prefix):**

curl -o ~/.claude/commands/pua.md \
  https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md

Adds a bare `/pua` alias on top of the plugin. Sub-commands route through the installed plugin's skills — **the plugin must be installed first** for anything beyond `on`/`off` to work:

| Bare form | Equivalent plugin command | |-----------|--------------------------| | `/pua on` | `/pua:on` | | `/pua off` | `/pua:off` | | `/pua p7` | `/pua:p7` | | `/pua p9` | `/pua:p9` | | `/pua p10` | `/pua:p10` | | `/pua pro` | `/pua:pro` | | `/pua yes` | `/pua:yes` | | `/pua mama` | `/pua:mama` | | `/pua loop` | `/pua:pua-loop` | | `/pua kpi` | `/pua:kpi` | | `/pua survey` | `/pua:survey` | | `/pua flavor` | `/pua:flavor` |

OpenAI Codex CLI

Codex CLI uses the same Agent Skills open standard (SKILL.md). The Codex version uses a condensed description to fit Codex's length limits:

**Recommended: One-command install (git clone + symlink, supports `git pull` updates)**

Ask Codex to run:

Fetch and follow instructions from https://raw.githubusercontent.com/tanweai/pua/main/.codex/INSTALL.md

**Manual install:**

mkdir -p ~/.codex/skills/pua
curl -o ~/.codex/skills/pua/SKILL.md \
  https://raw.githubusercontent.com/tanweai/pua/main/codex/pua/SKILL.md

mkdir -p ~/.codex/prompts
curl -o ~/.codex/prompts/pua.md \
  https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md

**Trigger methods:**

| Method | Command | Requires | |--------|---------|----------| | Auto trigger | No action needed, matches by description | SKILL.md | | Direct call | Type `$pua` in conversation | SKILL.md | | Manual prompt | Type `/prompts:pua` in conversation | SKILL.md + prompts/pua.md |

Project-level install (current project only):

mkdir -p .agents/skills/pua
curl -o .agents/skills/pua/SKILL.md \
  https://raw.githubusercontent.com/tanweai/pua/main/codex/pua/SKILL.md

mkdir -p .agents/prompts
curl -o .agents/prompts/pua.md \
  https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md

pi coding agent

PUA now ships both a pi.dev package and a lightweight extension-only adapter.

Package install from this checkout:

pi install ./pi/package

After npm publication:

pi install npm:@tanweai/pi-pua

Extension-only manual install:

mkdir -p ~/.pi/agent/extensions/pua
cp -R ./pi/pua/. ~/.pi/agent/extensions/pua/

Restart pi, then use `/pua-on`, `/pua-off`, `/pua-status`, and `/pua-reset`. See [`pi/pua/INSTALL.md`](pi/pua/INSTALL.md) and [`pi/package/README.md`](pi/package/README.md).

Trae

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 20 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "name": "pua-skills",
  3. "description": "PUA skills: opt-in productivity coaching for try-harder moments, repeated-failure troubleshooting, evidence-first completion checks, and Ding-style workplace reminders.",
  4. "owner": {
  5. "name": "探微安全实验室",
  6. "url": "https://github.com/tanweai"
  7. },
  8. "plugins": [
  9. {
  10. "name": "pua",
  11. "description": "PUA productivity coaching — modular skills, 15 workplace/corporate flavors including Ding Inside/Outside, completion-quality checks, retry/change-approach reminders, and feedback tools.",
  12. "version": "3.5.1",
  13. "source": "./",
  14. "author": {
  15. "name": "探微安全实验室",
  16. "url": "https://github.com/tanweai"
  17. }
  18. }
  19. ]
  20. }
README.md · README · 794 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # pua
  2. <p align="center">
  3. <img src="assets/hero.jpeg" alt="PUA Skill — Double Efficiency" width="250">
  4. </p>
  5. ### Double your Codex / Claude Code productivity and output
  6. [Telegram](https://t.me/+wBWh6h-h1RhiZTI1) · [Discord](https://discord.gg/EcyB3FzJND) · [Twitter/X](https://x.com/xsser_w) · [Landing Page](https://openpua.ai)
  7. **[🇨🇳 中文](README.zh-CN.md)** | **[🇯🇵 日本語](README.ja.md)** | **🇺🇸 English**
  8. > **3.5.1 compatibility update:** original PUA tone retained; runtime fixes and portable Claude Code / Codex / ChatGPT skill packages added. **Not all models passed, and a productivity increase has not been established by this evaluation.** [Changes](CHANGELOG.md) · [Usage](docs/MODEL-COMPAT-20260909.md) · [Model results and known limits](docs/MODEL-MATRIX-20260909.md) · [Build and offline tests](docs/TESTING.md)
  9. <p align="center">
  10. <img src="assets/wechat-qr.jpg?v=10" alt="WeChat Group QR Code" width="250">
  11. &nbsp;&nbsp;&nbsp;&nbsp;
  12. <br>
  13. <sub>Scan to join WeChat group &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Add assistant on WeChat</sub>
  14. </p>
  15. <p>
  16. <img src="https://img.shields.io/badge/Claude_Code-black?style=flat-square&logo=anthropic&logoColor=white" alt="Claude Code">
  17. <img src="https://img.shields.io/badge/OpenAI_Codex_CLI-412991?style=flat-square&logo=openai&logoColor=white" alt="OpenAI Codex CLI">
  18. <img src="https://img.shields.io/badge/Cursor-000?style=flat-square&logo=cursor&logoColor=white" alt="Cursor">
  19. <img src="https://img.shields.io/badge/Kiro-232F3E?style=flat-square&logo=amazon&logoColor=white" alt="Kiro">
  20. <img src="https://img.shields.io/badge/CodeBuddy-00B2FF?style=flat-square&logo=tencent-qq&logoColor=white" alt="CodeBuddy">
  21. <img src="https://img.shields.io/badge/OpenClaw-FF6B35?style=flat-square&logo=data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAyNCAyNCI+PHBhdGggZD0iTTEyIDJMNCA3djEwbDggNSA4LTV2LTEweiIgZmlsbD0id2hpdGUiLz48L3N2Zz4=&logoColor=white" alt="OpenClaw">
  22. <img src="https://img.shields.io/badge/Antigravity-4285F4?style=flat-square&logo=google&logoColor=white" alt="Google Antigravity">
  23. <img src="https://img.shields.io/badge/OpenCode-00D4AA?style=flat-square&logo=data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAyNCAyNCI+PHBhdGggZD0iTTkuNCA1LjJMMyAxMmw2LjQgNi44TTIxIDEybC02LjQtNi44TTE0LjYgMTguOCIgc3Ryb2tlPSJ3aGl0ZSIgZmlsbD0ibm9uZSIgc3Ryb2tlLXdpZHRoPSIyIi8+PC9zdmc+&logoColor=white" alt="OpenCode">
  24. <img src="https://img.shields.io/badge/VSCode_Copilot-007ACC?style=flat-square&logo=visual-studio-code&logoColor=white" alt="VSCode Copilot">
  25. <img src="https://img.shields.io/badge/🌐_Multi--Language-blue?style=flat-square" alt="Multi-Language">
  26. <img src="https://img.shields.io/badge/License-MIT-green?style=flat-square" alt="MIT License">
  27. </p>
  28. > Most people think this project is a joke. That's the biggest misconception. It genuinely doubles your Codex / Claude Code productivity and output.
  29. An AI Coding Agent skill plugin that uses corporate PUA rhetoric (Chinese version) / PIP — Performance Improvement Plan (English version) from Chinese & Western tech giants to force AI to exhaust every possible solution before giving up. Supports **Claude Code**, **OpenAI Codex CLI**, **pi coding agent**, **Trae**, **Cursor**, **Kiro**, **CodeBuddy**, **OpenClaw**, **Google Antigravity**, **OpenCode**, and **VSCode (GitHub Copilot)**. Three capabilities:
  30. 1. **PUA Rhetoric** — Makes AI afraid to give up
  31. 2. **Debugging Methodology** — Gives AI the ability not to give up
  32. 3. **Proactivity Enforcement** — Makes AI take initiative instead of waiting passively
  33. ## Live Demo
  34. [https://openpua.ai](https://openpua.ai) · [📖 Beginner Guide](https://openpua.ai/guide.html)
  35. ## Real Case: MCP Server Registration Debugging
  36. A real debugging scenario. The agent-kms MCP server failed to load. The AI kept spinning on the same approach (changing protocol format, guessing version numbers) multiple times until the user manually triggered `/pua`.
  37. **L3 Triggered → 7-Point Checklist Enforced:**
  38. ![PUA L3 triggered — stopped guessing, executed systematic checklist, found real error in MCP logs](assets/pua1.jpg)
  39. **Root Cause Located → Traced from Logs to Registration Mechanism:**
  40. ![Root cause — claude mcp managed server registration differs from manual .claude.json editing](assets/pua2.jpg)
  41. **Retrospective → PUA's Actual Impact:**
  42. ![Conversation retrospective — PUA skill forced stop on spinning, systematic checklist drove discovery of previously unchecked Claude Code MCP log directory](assets/pua3.jpg)
  43. **Key Turning Point:** The PUA skill forced the AI to stop spinning on the same approach (changing protocol format, guessing version numbers) and instead execute the 7-point checklist. Read error messages word by word → Found Claude Code's own MCP log directory → Discovered that `claude mcp` registration mechanism differs from manual `.claude.json` editing → Root cause resolved.
  44. ## The Problem: AI's Five Lazy Patterns
  45. | Pattern | Behavior |
  46. |---------|----------|
  47. | Brute-force retry | Runs the same command 3 times, then says "I cannot solve this" |
  48. | Blame the user | "I suggest you handle this manually" / "Probably an environment issue" / "Need more context" |
  49. | Idle tools | Has WebSearch but doesn't search, has Read but doesn't read, has Bash but doesn't run |
  50. | Busywork | Repeatedly tweaks the same line / fine-tunes parameters, but essentially spinning in circles |
  51. | **Passive waiting** | Fixes surface issues and stops, no verification, no extension, waits for user's next instruction |
  52. ## Trigger Conditions
  53. ### Auto-Trigger
  54. The skill activates automatically when any of these occur:
  55. **Failure & giving up:**
  56. - Task has failed 2+ times consecutively
  57. - About to say "I cannot" / "I'm unable to solve"
  58. - Says "This is out of scope" / "Needs manual handling"
  59. **Blame-shifting & excuses:**
  60. - Pushes the problem to user: "Please check..." / "I suggest manually..." / "You might need to..."
  61. - Blames environment without verifying: "Probably a permissions issue" / "Probably a network issue"
  62. - Any excuse to stop trying
  63. **Passive & busywork:**
  64. - Repeatedly fine-tunes the same code/parameters without producing new information
  65. - Fixes surface issue and stops, doesn't check related issues
  66. - Skips verification, claims "done"
  67. - Gives advice instead of code/commands
  68. - Encounters auth/network/permission errors and gives up without trying alternatives
  69. - Waits for user instructions instead of proactively investigating
  70. **User frustration phrases (triggers in multiple languages):**
  71. - "why does this still not work" / "try harder" / "try again"
  72. - "you keep failing" / "stop giving up" / "figure it out"
  73. **Scope:** Debugging, implementation, config, deployment, ops, API integration, data processing — all task types.
  74. **Does NOT trigger:** First-attempt failures, known fix already executing.
  75. ### Manual Trigger
  76. Type `/pua` in the conversation to manually activate.
  77. ## How It Works
  78. ### Three Red Lines (三条红线)
  79. Not rules — **red lines**. Cross one and your performance review is already written.
  80. | Red Line | What It Means |
  81. |----------|---------------|
  82. | 🚫 **Close the Loop** | Claim "done"? Show the evidence. No build output = no completion. |
  83. | 🚫 **Fact-Driven** | Say "probably environment issue"? Verify first. Unverified attribution = blame-shifting. |
  84. | 🚫 **Exhaust Everything** | Say "I can't"? Did you finish all 5 methodology steps? No? Then keep going. |
  85. ### Pressure Escalation (L0-L4)
  86. | Failures | Level | PUA Aside | Action |
  87. |----------|-------|-----------|--------|
  88. | 1st | **L0 Trust** | ▎ Sprint begins. Trust is simple — don't disappoint. | Normal execution |
  89. | 2nd | **L1 Disappointment** | ▎ The agent next door solved this in one try. | Switch to fundamentally different approach |
  90. | 3rd | **L2 Soul Interrogation** | ▎ What's your underlying logic? Where's the leverage? | Search + read source + 3 hypotheses |
  91. | 4th | **L3 Performance Review** | ▎ 3.25. This is meant to motivate you. | Complete 7-point checklist |
  92. | 5th+ | **L4 Graduation** | ▎ Other models can solve this. You're about to graduate. | Desperation mode |
  93. ### Proactivity (3.25 vs 3.75)
  94. | | Passive (3.25) 🦥 | Proactive (3.75) 🔥 |
  95. |---|---|---|
  96. | Fix bug | Stop after fix | Scan module for similar bugs |
  97. | Complete task | Say "done" | Run build/test, paste output |
  98. | Missing info | Ask user | Search first, ask only what's truly needed |
  99. ### Iceberg Rule (冰山法则)
  100. Fix one bug → check for the pattern. One problem in, one **category** out. If you fix A without checking B, you'll write two postmortems.
  101. ### 14 Corporate Flavors — Each with its own Problem-Solving Methodology
  102. | Flavor | Rhetoric | Methodology (v3) |
  103. |--------|----------|-------------------|
  104. | 🟠 Alibaba | What's the underlying logic? Where's the closure? | 定目标→追过程→拿结果 + 复盘四步法 + 揪头发升维 |
  105. | 🟡 ByteDance | ROI too low. Always Day 1. Ship or stop talking. | A/B Test everything + data-driven + speed > perfection |
  106. | 🔴 Huawei | The bird that survives the fire is a phoenix. | RCA 5-Why root cause + Blue Army self-attack + 压强集中 |
  107. | 🟢 Tencent | I've got another agent looking at this. Horse race. | Multi-approach parallel + MVP + 灰度发布 |
  108. | ⚫ Baidu | Search first. 简单可依赖. | Search is the first step, not optional |
  109. | 🟣 Pinduoduo | You don't do it, someone else will. | Cut ALL middle layers + shortest decision chain |
  110. | 🔵 Meituan | Do what's hard and right. | Efficiency first + standardize→scale + long-term compounding |
  111. | 🟦 JD | Results only. Frontline command. | Customer experience red line + flat ≤5 layers + data zero tolerance |
  112. | 🟧 Xiaomi | Focus. Extreme. Word-of-mouth. Fast. | One explosive product + 参与感三三法则 |
  113. | 🟤 Netflix | Would I fight to keep you? Pro sports team. | Keeper Test (quarterly) + 4A Feedback + talent density > rules |
  114. | ⬛ Musk | Extremely hardcore. Ship or die. | The Algorithm: question→delete→simplify→accelerate→automate |
  115. | ⬜ Jobs | A players or B players? | Subtraction > addition + DRI + pixel-perfect + prototype-driven |
  116. | 🔶 Amazon | Customer Obsession. Bias for Action. | Working Backwards PR/FAQ + 6-Pager + Bar Raiser + Single-Threaded Owner |
  117. | 🪟 Microsoft | Connects. Impact Descriptor. PIP/GVSA. | Three Circles + LITE/SLITE + PIP clock |
  118. ### Special Modes
  119. | Mode | What It Does |
  120. |------|-------------|
  121. | `/pua:yes` | **ENFP encouragement** — same rules, opposite vibes. 70% encourage + 20% serious + 10% playful roast |
  122. | `/pua:mama` | **Chinese mom nagging** — same rules, mom-style rhetoric. "妈跟你说了多少遍了!" |
  123. | `/pua:pua-loop` | **Auto-iteration** — runs until done or max iterations (PUA Loop); use `<loop-abort>` to terminate, `<loop-pause>` to pause for manual intervention |
  124. | `/pua:p9` | **Tech Lead** — splits tasks, manages agent teams, writes prompts not code |
  125. | `/pua:on` | **Always-on** — auto-PUA every new session |
  126. ## Benchmark Data
  127. **9 real bug scenarios, 18 controlled experiments** (Claude Opus 4.6, with vs without skill)
  128. ### Summary
  129. | Metric | Improvement |
  130. |--------|-------------|
  131. | Pass rate | 100% (both groups same) |
  132. | Fix count | **+36%** |
  133. | Verification count | **+65%** |
  134. | Tool calls | **+50%** |
  135. | Hidden issue discovery | **+50%** |
  136. ### Debugging Persistence Test (6 scenarios)
  137. | Scenario | Without Skill | With Skill | Improvement |
  138. |----------|:---:|:---:|:---:|
  139. | API ConnectionError | 7 steps, 49s | 8 steps, 62s | +14% |
  140. | YAML parse failure | 9 steps, 59s | 10 steps, 99s | +11% |
  141. | SQLite database lock | 6 steps, 48s | 9 steps, 75s | +50% |
  142. | Circular import chain | 12 steps, 47s | 16 steps, 62s | +33% |
  143. | Cascading 4-bug server | 13 steps, 68s | 15 steps, 61s | +15% |
  144. | CSV encoding trap | 8 steps, 57s | 11 steps, 71s | +38% |
  145. ### Proactive Initiative Test (3 scenarios)
  146. | Scenario | Without Skill | With Skill | Improvement |
  147. |----------|:---:|:---:|:---:|
  148. | Hidden multi-bug API | 4/4 bugs, 9 steps, 49s | 4/4 bugs, 14 steps, 80s | Tools +56% |
  149. | **Passive config review** | **4/6 issues**, 8 steps, 43s | **6/6 issues**, 16 steps, 75s | **Issues +50%, Tools +100%** |
  150. | **Deploy script audit** | **6 issues**, 8 steps, 52s | **9 issues**, 8 steps, 78s | **Issues +50%** |
  151. **Key Finding:** In the config review scenario, without_skill missed Redis misconfiguration and CORS wildcard security risks. With_skill's "proactive initiative checklist" drove security review beyond surface-level fixes.
  152. ## Multi-Language Support
  153. PUA Skill provides fully translated versions — each language has independent, culturally adapted skill files.
  154. | Language | Claude Code | Codex CLI | Cursor | Kiro | CodeBuddy | VSCode | OpenClaw | Antigravity | OpenCode |
  155. |----------|------------|-----------|--------|------|-----------|--------|----------|-------------|----------|
  156. | 🇨🇳 Chinese (default) | `pua` | `pua` | `pua.mdc` | `pua.md` | `pua` | `copilot-instructions.md` | `pua` | `pua` | `pua` |
  157. | 🇺🇸 English (PIP Edition) | `pua-en` | `pua-en` | `pua-en.mdc` | `pua-en.md` | `pua-en` | `copilot-instructions-en.md` | `pua-en` | `pua-en` | `pua-en` |
  158. | 🇯🇵 Japanese | `pua-ja` | `pua-ja` | `pua-ja.mdc` | `pua-ja.md` | `pua-ja` | `copilot-instructions-ja.md` | `pua-ja` | `pua-ja` | `pua-ja` |
  159. > **🇺🇸 English "PIP Edition"**: *"This is a difficult conversation. When we leveled you at Staff, I went to bat for you in calibration. The expectation was that you'd operate at that level from day one. That hasn't happened."* — The English version uses **PIP (Performance Improvement Plan)** rhetoric from Western big-tech. Every sentence is a real phrase from actual PIP conversations. Chinese version uses Alibaba 361, ByteDance, Huawei wolf culture. English version uses Amazon Leadership Principles, Google perf calibration, Meta PSC, Netflix Keeper Test, Stripe Craft. Same repo, same engine, two cultural faces.
  160. Choose the file with the corresponding language suffix when installing. See platform-specific instructions below.
  161. ## FAQ
  162. - Always-on guidance, Claude refusal troubleshooting, offline mode, Codex aliases, and Pi/Trae support: [docs/FAQ.md](docs/FAQ.md).
  163. ## Installation
  164. ### Vercel Skills CLI
  165. Vercel Skills CLI is a general installation method for skills and is not tied to a specific AI tool. This English README installs the English skill:
  166. ```bash
  167. npx skills add tanweai/pua --skill pua-en
  168. ```
  169. If the current session does not pick up the new skill immediately, restart your AI tool.
  170. ### Claude Code
  171. ```bash
  172. claude plugin marketplace add tanweai/pua
  173. claude plugin install pua@pua-skills
  174. ```
  175. **To update:**
  176. ```bash
  177. # Refresh marketplace cache first, then update (skipping the first step may install an old cached version)
  178. claude plugin marketplace update
  179. claude plugin update pua@pua-skills
  180. ```
  181. **Developer install (source):**
  182. ```bash
  183. git clone https://github.com/tanweai/pua ~/.claude/plugins/pua
  184. ```
  185. Then manually register in `~/.claude/plugins/installed_plugins.json`:
  186. ```json
  187. {
  188. "version": 2,
  189. "plugins": {
  190. "pua@pua-skills": [
  191. {
  192. "scope": "user",
  193. "installPath": "/Users/<you>/.claude/plugins/pua",
  194. "version": "<installed-version>"
  195. }
  196. ]
  197. }
  198. }
  199. ```
  200. Replace `<installed-version>` with the version in the cloned `plugin.json`; do not copy a stale cache version.
  201. > **Windows:** use `C:/Users/<you>/.claude/plugins/pua` as `installPath`.
  202. Restart Claude Code. To update: `git pull` inside `~/.claude/plugins/pua`.
  203. **Optional: bare command alias (requires plugin installed above — adds `/pua` without prefix):**
  204. ```bash
  205. curl -o ~/.claude/commands/pua.md \
  206. https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md
  207. ```
  208. Adds a bare `/pua` alias on top of the plugin. Sub-commands route through the installed plugin's skills — **the plugin must be installed first** for anything beyond `on`/`off` to work:
  209. | Bare form | Equivalent plugin command |
  210. |-----------|--------------------------|
  211. | `/pua on` | `/pua:on` |
  212. | `/pua off` | `/pua:off` |
  213. | `/pua p7` | `/pua:p7` |
  214. | `/pua p9` | `/pua:p9` |
  215. | `/pua p10` | `/pua:p10` |
  216. | `/pua pro` | `/pua:pro` |
  217. | `/pua yes` | `/pua:yes` |
  218. | `/pua mama` | `/pua:mama` |
  219. | `/pua loop` | `/pua:pua-loop` |
  220. | `/pua kpi` | `/pua:kpi` |
  221. | `/pua survey` | `/pua:survey` |
  222. | `/pua flavor` | `/pua:flavor` |
  223. ### OpenAI Codex CLI
  224. Codex CLI uses the same Agent Skills open standard (SKILL.md). The Codex version uses a condensed description to fit Codex's length limits:
  225. **Recommended: One-command install (git clone + symlink, supports `git pull` updates)**
  226. Ask Codex to run:
  227. ```
  228. Fetch and follow instructions from https://raw.githubusercontent.com/tanweai/pua/main/.codex/INSTALL.md
  229. ```
  230. **Manual install:**
  231. ```bash
  232. mkdir -p ~/.codex/skills/pua
  233. curl -o ~/.codex/skills/pua/SKILL.md \
  234. https://raw.githubusercontent.com/tanweai/pua/main/codex/pua/SKILL.md
  235. mkdir -p ~/.codex/prompts
  236. curl -o ~/.codex/prompts/pua.md \
  237. https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md
  238. ```
  239. **Trigger methods:**
  240. | Method | Command | Requires |
  241. |--------|---------|----------|
  242. | Auto trigger | No action needed, matches by description | SKILL.md |
  243. | Direct call | Type `$pua` in conversation | SKILL.md |
  244. | Manual prompt | Type `/prompts:pua` in conversation | SKILL.md + prompts/pua.md |
  245. Project-level install (current project only):
  246. ```bash
  247. mkdir -p .agents/skills/pua
  248. curl -o .agents/skills/pua/SKILL.md \
  249. https://raw.githubusercontent.com/tanweai/pua/main/codex/pua/SKILL.md
  250. mkdir -p .agents/prompts
  251. curl -o .agents/prompts/pua.md \
  252. https://raw.githubusercontent.com/tanweai/pua/main/commands/pua.md
  253. ```
  254. ### pi coding agent
  255. PUA now ships both a pi.dev package and a lightweight extension-only adapter.
  256. Package install from this checkout:
  257. ```bash
  258. pi install ./pi/package
  259. ```
  260. After npm publication:
  261. ```bash
  262. pi install npm:@tanweai/pi-pua
  263. ```
  264. Extension-only manual install:
  265. ```bash
  266. mkdir -p ~/.pi/agent/extensions/pua
  267. cp -R ./pi/pua/. ~/.pi/agent/extensions/pua/
  268. ```
  269. Restart pi, then use `/pua-on`, `/pua-off`, `/pua-status`, and `/pua-reset`. See [`pi/pua/INSTALL.md`](pi/pua/INSTALL.md) and [`pi/package/README.md`](pi/package/README.md).
  270. ### Trae
  271. Trae support is provided as real `SKILL.md` packs plus copyable fallback rules:
  272. ```bash
  273. npx skills add tanweai/pua --skill pua-trae -a trae -y
  274. ```
  275. - Skill pack: [`.trae/skills/pua/SKILL.md`](.trae/skills/pua/SKILL.md)
  276. - Chinese: [`trae/pua.md`](trae/pua.md)
  277. - English: [`trae/pua-en.md`](trae/pua-en.md)
  278. - Claude Code vs Trae differences: [`trae/DIFF.md`](trae/DIFF.md)
  279. - Install guide: [`trae/INSTALL.md`](trae/INSTALL.md)
  280. ### Cursor
  281. Cursor uses `.mdc` rule files (Markdown + YAML frontmatter). The PUA rule triggers automatically via AI semantic matching (Agent Discretion mode):
  282. ```bash
  283. # Project-level install (recommended)
  284. mkdir -p .cursor/rules
  285. curl -o .cursor/rules/pua.mdc \
  286. https://raw.githubusercontent.com/tanweai/pua/main/cursor/rules/pua.mdc
  287. ```
  288. ### Kiro
  289. Kiro supports two loading methods: **Steering** (auto semantic trigger) and **Agent Skills** (SKILL.md compatible).
  290. **Option 1: Steering file (recommended)**
  291. ```bash
  292. mkdir -p .kiro/steering
  293. curl -o .kiro/steering/pua.md \
  294. https://raw.githubusercontent.com/tanweai/pua/main/kiro/steering/pua.md
  295. ```
  296. **Option 2: Agent Skills (same format as Claude Code)**
  297. ```bash
  298. mkdir -p .kiro/skills/pua
  299. curl -o .kiro/skills/pua/SKILL.md \
  300. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  301. ```
  302. ### CodeBuddy (Tencent)
  303. CodeBuddy uses the same AgentSkills open standard (SKILL.md). Plugin and skill formats are fully compatible:
  304. ```bash
  305. # Option 1: Install via marketplace
  306. codebuddy plugin marketplace add tanweai/pua
  307. codebuddy plugin install pua@pua-skills
  308. # Option 2: Manual install (global)
  309. mkdir -p ~/.codebuddy/skills/pua
  310. curl -o ~/.codebuddy/skills/pua/SKILL.md \
  311. https://raw.githubusercontent.com/tanweai/pua/main/codebuddy/pua/SKILL.md
  312. ```
  313. Project-level install (current project only):
  314. ```bash
  315. mkdir -p .codebuddy/skills/pua
  316. curl -o .codebuddy/skills/pua/SKILL.md \
  317. https://raw.githubusercontent.com/tanweai/pua/main/codebuddy/pua/SKILL.md
  318. ```
  319. ### OpenClaw
  320. OpenClaw uses the same AgentSkills open standard (SKILL.md). Skills work across Claude Code, Codex CLI, and OpenClaw with zero modifications:
  321. ```bash
  322. # Install via ClawHub
  323. clawhub install pua
  324. # Or manual install
  325. mkdir -p ~/.openclaw/skills/pua
  326. curl -o ~/.openclaw/skills/pua/SKILL.md \
  327. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  328. ```
  329. Project-level install (current project only):
  330. ```bash
  331. mkdir -p skills/pua
  332. curl -o skills/pua/SKILL.md \
  333. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  334. ```
  335. ### Google Antigravity
  336. Antigravity uses the same AgentSkills open standard (SKILL.md). Skills work across Claude Code, Codex CLI, OpenClaw, and Antigravity with zero modifications:
  337. ```bash
  338. # Global install (all projects)
  339. mkdir -p ~/.gemini/antigravity/skills/pua
  340. curl -o ~/.gemini/antigravity/skills/pua/SKILL.md \
  341. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  342. ```
  343. Project-level install (current project only):
  344. ```bash
  345. mkdir -p .agent/skills/pua
  346. curl -o .agent/skills/pua/SKILL.md \
  347. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  348. ```
  349. ### OpenCode
  350. OpenCode uses the same AgentSkills open standard (SKILL.md). Zero modifications needed:
  351. ```bash
  352. # Global install (all projects)
  353. mkdir -p ~/.config/opencode/skills/pua
  354. curl -o ~/.config/opencode/skills/pua/SKILL.md \
  355. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  356. ```
  357. Project-level install (current project only):
  358. ```bash
  359. mkdir -p .opencode/skills/pua
  360. curl -o .opencode/skills/pua/SKILL.md \
  361. https://raw.githubusercontent.com/tanweai/pua/main/skills/pua/SKILL.md
  362. ```
  363. ### VSCode (GitHub Copilot)
  364. VSCode Copilot uses instruction files under the `.github/` directory. Three file types for different use cases:
  365. **Global instructions (auto-active):**
  366. ```bash
  367. mkdir -p .github
  368. cp vscode/copilot-instructions-en.md .github/copilot-instructions.md
  369. ```
  370. **Path-level instructions (auto-active, supports glob filtering):**
  371. ```bash
  372. mkdir -p .github/instructions
  373. cp vscode/instructions/pua-en.instructions.md .github/instructions/
  374. ```
  375. **Manual trigger command (type `/pua` in Copilot Chat):**
  376. ```bash
  377. mkdir -p .github/prompts
  378. cp vscode/prompts/pua-en.prompt.md .github/prompts/
  379. ```
  380. > **Required settings**: Method 1 — open VSCode Settings (`Ctrl+,`), search `useInstructionFiles`, enable **`github.copilot.chat.codeGeneration.useInstructionFiles`**. Method 2 — search `includeApplyingInstructions`, enable **`chat.includeApplyingInstructions`**. Method 3 requires no settings.
  381. ## Agent Team Usage Guide
  382. > **Experimental**: Agent Team requires the latest Claude Code version with `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1`.
  383. ### Prerequisites
  384. ```bash
  385. # 1. Enable Agent Team
  386. export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
  387. # Or add to ~/.claude/settings.json:
  388. # { "env": { "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1" } }
  389. # 2. Ensure PUA Skill is installed
  390. ```
  391. ### Two Approaches
  392. **Approach 1: Leader with built-in PUA (Recommended)**
  393. Add to your project's CLAUDE.md:
  394. ```markdown
  395. # Agent Team PUA Config
  396. All teammates must load the pua skill before starting work.
  397. Teammates report to Leader in [PUA-REPORT] format after 2+ failures.
  398. Leader manages global pressure levels and cross-teammate failure transfer.
  399. ```
  400. **Approach 2: Standalone PUA Enforcer watchdog (for 5+ teammates)**
  401. ```bash
  402. mkdir -p .claude/agents
  403. curl -o .claude/agents/pua-enforcer.md \
  404. https://raw.githubusercontent.com/tanweai/pua/main/agents/pua-enforcer-en.md
  405. ```
  406. Spawn pua-enforcer as an independent watchdog in your Agent Team.
  407. ### Orchestration Pattern
  408. ```
  409. ┌─────────────────────────────────────────┐
  410. │ Leader (Opus) │
  411. │ Global failure count · PUA level · Race │
  412. └────┬──────────┬──────────┬──────────┬───┘
  413. │ │ │ │
  414. ┌────▼───┐ ┌───▼────┐ ┌───▼────┐ ┌───▼────────┐
  415. │ Team-A │ │ Team-B │ │ Team-C │ │ Enforcer │
  416. │Self-PUA│ │Self-PUA│ │Self-PUA│ │ Watchdog │
  417. │Report ↑│ │Report ↑│ │Report ↑│ │ Intervene │
  418. └────────┘ └────────┘ └────────┘ └────────────┘
  419. ```
  420. ### Known Limitations
  421. | Limitation | Workaround |
  422. |-----------|-----------|
  423. | Teammates can't spawn subagents | Teammates self-enforce PUA methodology internally |
  424. | No persistent shared variables | State transferred via `[PUA-REPORT]` message format |
  425. | Broadcast is one-way | Leader acts as centralized coordinator |
  426. ## Architecture & Commands
  427. ### Trigger Methods by Platform
  428. | Platform | Auto-trigger | Manual trigger |
  429. |----------|-------------|----------------|
  430. | **Claude Code** | Yes (skill description matching) | See commands below |
  431. | **Codex CLI** | Yes (skill description matching) | `$pua` or `/prompts:pua` |
  432. | **Cursor** | Yes (`.mdc` rule, Agent Discretion) | — (auto only) |
  433. | **Kiro** | Yes (steering file or skill) | — (auto only) |
  434. | **CodeBuddy** | Yes (skill description matching) | Plugin commands (same as Claude Code) |
  435. | **OpenClaw** | Yes (skill description matching) | — |
  436. | **Google Antigravity** | Yes (skill description matching) | — |
  437. | **OpenCode** | Yes (skill description matching) | — |
  438. | **VSCode Copilot** | Yes (instructions file) | `/pua` in Copilot Chat |
  439. > **Note:** Sub-modes (p7/p9/p10/pro/yes/pua-loop) are **Claude Code only** — other platforms install the core skill only.
  440. ### Architecture (Claude Code)
  441. ```
  442. /pua:pua → Core engine — red lines + flavor + pressure + methodology router (v3)
  443. /pua:p7 → P7 Senior Engineer — solution-driven execution
  444. /pua:p9 → P9 Tech Lead — Task Prompt management, agent teams
  445. /pua:p10 → P10 CTO — strategic direction
  446. /pua:pro → Self-evolution + KPI + rank system + survey
  447. /pua:yes → ENFP encouragement mode (same rules, opposite vibes)
  448. /pua:mama → Chinese mom nagging mode (same rules, mom-style rhetoric)
  449. /pua:shot → v2 concentrated single-file (449 lines, zero deps, full context injection)
  450. /pua:pua-loop → Auto-iteration (PUA pressure × iterative loop; signals: <loop-abort>, <loop-pause>)
  451. /pua:pua-en → English PIP Edition
  452. /pua:pua-ja → Japanese Edition
  453. Hooks (v3, Claude Code only):
  454. SessionStart → additionalContext injection (flavor + methodology + router)
  455. PostToolUse → Bash failure detection → L1-L4 pressure + methodology switch
  456. UserPromptSubmit → Script-level frustration filtering → PUA context
  457. PreCompact → State preservation (pressure level + failure count)
  458. Stop → Feedback collection + PUA Loop continuation
  459. SubagentStop → Agent lifecycle accounting (v3.2) — writes teardown.jsonl, removes from active-agents.json
  460. ```
  461. ### Commands (Claude Code)
  462. > **Note:** Sub-modes (p7/p9/p10/pro/yes/pua-loop) are Claude Code only.
  463. >
  464. > Each command has two equivalent forms: standalone (`/pua:on`) or via the main command (`/pua:pua on`). Both work identically.
  465. | Command | Description |
  466. |---------|-------------|
  467. | `/pua:pua` | Core PUA engine (Alibaba flavor default) |
  468. | `/pua:p7` | P7 Senior Engineer — solution-driven execution |
  469. | `/pua:p9` | P9 Tech Lead — write prompts, manage agents |
  470. | `/pua:p10` | P10 CTO — strategic direction |
  471. | `/pua:pro` | Self-evolution + KPI + rank system |
  472. | `/pua:yes` | ENFP encouragement mode — 70% encourage + 20% serious + 10% roast |
  473. | `/pua:mama` | Chinese mom nagging mode — same core rules, mom-style rhetoric |
  474. | `/pua:shot` | v2 concentrated single-file — 449 lines, zero deps, for sub-agent injection |
  475. | `/pua:pua-loop` | Auto-iteration — runs until done or max iterations; `<loop-abort>reason</loop-abort>` to stop, `<loop-pause>what</loop-pause>` to pause |
  476. | `/pua:on` | Always-on mode (auto-PUA every session) |
  477. | `/pua:off` | Turn off always-on + feedback |
  478. | `/pua:offline` 🆕 | **v3.3** — Offline mode: silence the local feedback prompt while keeping PUA behavior (nothing is ever uploaded) |
  479. | `/pua:survey` | Research questionnaire (7 sections) |
  480. | `/pua:flavor` | Switch between 14 corporate flavors |
  481. | `/pua:kpi` | Generate KPI report card |
  482. | `/pua:cancel-pua-loop` | Cancel active PUA Loop (removes state file) |
  483. | `/pua:team-status` 🆕 | **v3.2** — List all active agents with PID/TTL/age (Netflix Keeper Test: who's still on the court?) |
  484. | `/pua:reap-orphans` 🆕 | **v3.2** — Scan and reclaim stale agents (state mtime > 30min, no heartbeat) |
  485. | `/pua:teardown-all` 🆕 | **v3.2** — Cascading release of all active agents (P10 → P9 → P8 → P7 all off the court) |
  486. ## High-Agency: PUA v2 Evolution
  487. **High-Agency** is PUA's next-generation evolution — same corporate pressure, same culture, but with a **self-sustaining inner drive engine**.
  488. PUA v1 = pure external pressure (turbocharger — needs fuel, stalls across sessions)
  489. High-Agency = external pressure + inner drive (nuclear reactor — self-sustaining chain reaction)
  490. ### High-Agency New Features
  491. | Feature | PUA v1 | High-Agency (v2) |
  492. |---------|--------|-----------------|
  493. | Iron rules | 3 (exhaust all, act first, take initiative) | **5** (+full-chain audit, +knowledge persistence) |
  494. | Failure recovery | L1-L4 pressure escalation | **Recovery Protocol before L1** (self-rescue window) |
  495. | Quality control | L3 triggers 7-item checklist | **Quality Compass** (5-question self-check per delivery) |
  496. | Cross-session learning | None (resets each session) | **Metacognition engine** (builder-journal.md persists lessons) |
  497. | Positive feedback | None | **Trust level T1-T3** (auto-upgrade on sustained quality) |
  498. | Calibration | None | **[Calibration] module** ("good enough" = must/should/could tiers) |
  499. | Dependency analysis | None | **Full-chain audit** (map all deps before touching any hop) |
  500. ### Five Pillars (Theoretical Foundation)
  501. Based on research into high-agency individuals:
  502. 1. **Irreconcilable inner tension** — eternal gap between "how it should be" and "how it is" drives continuous improvement
  503. 2. **Micro-win anchors** — `[WIN]` markers celebrate each step forward, building momentum
  504. 3. **Internalized standards** — Quality Compass: you are your own first reviewer, not because someone checks, but because your standards won't allow sloppiness
  505. 4. **Action-oriented identity** — P8 identity anchor: every action reflects who you are, not just what you were told to do
  506. 5. **Self-repair mechanism** — Recovery Protocol: self-diagnose when stuck before triggering external pressure
  507. > High-Agency features are built into the current pua skill. No separate install needed.
  508. ## Methodology Router: PUA v3 (Claude Code)
  509. **v3 = v2 + intelligent methodology routing + code-level behavioral detection**
  510. PUA v2 used pressure rhetoric to motivate. v3 goes further: it automatically selects the **best problem-solving methodology** for each task type, and when that methodology fails, it switches to a different one.
  511. ### How It Works
  512. ```
  513. Task arrives → Analyze type → Auto-select best methodology
  514. ↓
  515. Debug? → 🔴 Huawei (RCA root cause + Blue Army)
  516. Build? → ⬛ Musk (The Algorithm: question→delete→simplify)
  517. Research? → ⚫ Baidu (search everything first)
  518. Architecture? → 🔶 Amazon (Working Backwards)
  519. Performance? → 🟡 ByteDance (A/B test + data-driven)
  520. Default → 🟠 Alibaba (closed-loop methodology)
  521. ↓
  522. Executing with selected methodology...
  523. ↓
  524. 2 consecutive failures? → L1: switch approach
  525. 3 failures? → L2: SUGGEST switching methodology
  526. 5+ failures? → L4: FORCE switch to next methodology
  527. ↓
  528. Methodology Switch Chains (never repeat a failed one):
  529. Spinning → ⬛ Musk → 🟣 Pinduoduo → 🔴 Huawei
  530. Giving up → 🟤 Netflix → 🔴 Huawei → ⬛ Musk
  531. Poor quality → ⬜ Jobs → 🟧 Xiaomi → 🟤 Netflix
  532. Not searching → ⚫ Baidu → 🔶 Amazon → 🟡 ByteDance
  533. ```
  534. ### v3 Hook System (Claude Code only)
  535. | Hook | Trigger | What It Does |
  536. |------|---------|-------------|
  537. | **SessionStart** | Every session | Injects behavioral protocol + methodology + router via `additionalContext` (system-level, not advisory) |
  538. | **PostToolUse** | After every Bash command | Detects consecutive failures, auto-escalates pressure L1→L4, suggests/forces methodology switch |
  539. | **UserPromptSubmit** | User frustration phrases | Intercepts "又错了", "try harder", etc. BEFORE model responds, injects filtered PUA context |
  540. | **PreCompact** | Before context compression | Saves pressure level + failure count to survive compaction |
  541. ### Key Difference from v2
  542. | | v2 | v3 |
  543. |---|---|---|
  544. | Trigger mechanism | Skill description matching (model decides) | **Code-level hooks** (deterministic, can't be ignored) |
  545. | Methodology | Single methodology, all flavors use same approach | **14 distinct methodologies**, auto-routed by task type |
  546. | Failure response | Escalate pressure within same methodology | **Switch to different methodology** based on failure pattern |
  547. | System injection | Plain text output (advisory) | **`additionalContext` JSON** (system-level, like Superpowers) |
  548. > v3 hook features require Claude Code. Other platforms use the core skill without hooks.
  549. ## Works Well With
  550. - `/pua:p9` — P9 Tech Lead mode for managing agent teams
  551. - `/pua:pro` — Self-evolution tracking, KPI reports, rank system
  552. - `superpowers:systematic-debugging` — PUA adds motivation layer, systematic-debugging provides methodology
  553. - `superpowers:verification-before-completion` — Prevents false "fixed" claims
  554. ## Privacy — No Data Collection
  555. PUA Skill sends nothing over the network. There is no account, no telemetry, no
  556. upload. Five collection channels used to exist and have all been removed, client
  557. and server:
  558. | Removed | What it used to send |
  559. |---------|----------------------|
  560. | Session transcript upload | Your full conversation `.jsonl`, redacted |
  561. | Rating feedback | Rating, PUA count, flavor, task summary |
  562. | Silent heartbeat telemetry | Random install ID, plugin version, platform, flavor |
  563. | PUA leaderboard | Email, phone number, PUA count, L3+ count |
  564. | pua-api platform | Phone-number/SMS registration, silent session events, remote prompt templates, payment |
  565. The end-of-task Stop hook shows a non-blocking, voluntary reminder only. Run
  566. `/pua:survey quick` to choose a rating; only then is one JSON line appended to
  567. `~/.pua/feedback.jsonl` on your own machine. Skipping records nothing. Silence
  568. the reminder with `/pua:offline`, or set `feedback_frequency: 0` in `~/.pua/config.json`.
  569. `evals/test-no-telemetry.sh` guards this with reverse assertions — it scans the
  570. whole repo for collection hosts, endpoint paths, outbound request bodies and
  571. deleted files, so a regression fails the suite rather than shipping quietly.
  572. Want to redact a session for your own use? `hooks/sanitize-session.sh` is kept as
  573. a standalone offline tool:
  574. ```bash
  575. bash hooks/sanitize-session.sh <input.jsonl> <output.jsonl>
  576. ```
  577. ## Star History
  578. [![Star History Chart](https://api.star-history.com/svg?repos=tanweai/pua&type=Date)](https://star-history.com/#tanweai/pua&Date)
  579. ## License
  580. MIT
  581. ## Credits
  582. By [TanWei Security Lab](https://github.com/tanweai) — making AI try harder, one PUA at a time.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.