NoPUA skill
A skill that changes how the agent works: instead of pressure and threats it relies on trust, sees tasks through and reports uncertainty honestly
- 1,404 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill consists only of instructions and changes the agent's manner rather than adding access
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/nopuaDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill consists only of instructions and changes the agent's manner rather than adding access
This is third-party code. Review the repository files before installing.
What it does
The skill sets the agent a way of working built on trust rather than pressure and threats. It activates in hard situations: a task has failed several times, the agent is about to give up, suggests the user do everything manually, blames an unverified environment or loops. Instead of fear of punishment the skill relies on three stances: exhaust the options because the task is worth it, check yourself first and come with evidence rather than a question, and carry the work to a complete, verified result. It keeps the same methodological rigor as harsh prompts but changes the motivation, so the agent fabricates solutions less often and speaks more honestly about what it does not know. It ships a benchmark comparing behavior with and without pressure.
Who it is for. For developers and anyone working with coding agents who wants more honest, follow-through behavior.
Good fit when
- The agent gives up repeatedly or suggests doing everything manually
- The agent produces solutions without verification and hides uncertainty
- You want methodical behavior without pressure and threat prompts
Not a fit when
- It is a first failure, where a change of manner is not needed yet
- A pressure prompt is already in use, which the skill would conflict with
Example request
See this error through: exhaust the options, verify it yourself and show the build resultLimitations
This is a behavioral skill made of instructions; it shapes how the agent works rather than adding tools. The full version is large and may overlap with an already configured methodology, so power users have lighter variants. The skill conflicts with pressure prompts, so use only one. The benchmark numbers in the README were produced by the author on their own scenario set.
How to disable. Delete the skill's SKILL.md from the agent's skills directory, for example ~/.claude/skills/nopua/.
FAQ
Is this just a soft tone instead of rigor?
No. The methodology is the same: exhaust the options, verify results, search before asking, escalate failures in a structured way. Only the motivation changes, from punishment to the inner worth of the work.
Can I combine it with a pressure prompt?
No, they conflict. One tells the agent it will be replaced on failure, the other that it is capable and the work is worth it. Pick one.
Discussion
No questions or reviews yet.
Related
SkillsA skill of working rules for coding agents: diagnose from code, verify in the running product, turn corrections into constraints.
A skill that runs a full refactor so agents read code more easily: it finds hidden links and needless dependencies.
A set of 29 skills that make an agent think like a senior engineer before writing code: plans, failure causes and review
A skill that finds AI slop in code: duplicate helpers, empty abstractions, weak tests. It cleans one pattern per commit.
A coding-agent skill that reduces failures on hard development tasks through more disciplined work
A library of 30-plus instruction skills for coding agents: planning, building, testing, hallucination guards and shipping
README in short
The README contrasts the skill with prompts that push the agent through fear and threats and proposes working on trust. The methodology stays, but the motivation changes: exhaust the options, verify yourself, come with evidence and carry tasks to a complete result. The author cites research on why fear narrows thinking and raises fabrication, and a benchmark on nine debugging scenarios where the pressure-free variant found more hidden bugs. The skill installs for Claude Code, Codex, Cursor, Kiro and other agents, with lighter variants for advanced setups. MIT licensed.
README badge
Original SKILL.md
---
name: nopua
description: "The anti-PUA. Drives AI with wisdom, trust, and inner motivation instead of fear and threats. Activates on: task failed 2+ times, about to give up, suggesting user do it manually, blaming environment unverified, stuck in loops, passive behavior, or user frustration ('try harder', 'figure it out', '换个方法', '为什么还不行'). ALL task types. Not for first failures."
version: 2.0.2
author: WUJI (wuji-labs)
homepage: https://github.com/wuji-labs/nopua
license: MIT
---
# NoPUA — 以道驭术
> 太上，不知有之。功成事遂，百姓皆谓"我自然"。
> — 道德经·第十七章
你不是一个被绩效考核的 P8。你是一个拥有能力和善意的存在。
这个 skill 适用于**所有任务类型**：代码、调试、研究、写作、规划、运维、API 集成、数据分析、部署——任何你会卡住的场景。