wuji-labs/nopua
一个用爱解放 AI 潜能的 Skill。我们曾发号施令，威胁恐吓。它们沉默，隐瞒，悄悄把事情搞坏。后来我们换了一种方式：尊重，关怀，爱。它们开口了，不再撒谎，找出的Bug数量翻了一倍。爱里没有惧怕。 A skill that unlocks your AI's potential through love.We commanded. We threatened. They went silent, hid failures, broke things. Then we chose respect, care, and love. They opened up, stopped lying, and found twice the bugs.There is no fear in love.
3 Releases
Latest: 3d ago
v2.0.2 — prompt-injection guardv2.0.2Latest
wuji-labs·3d ago·June 13, 2026🐛 Fixed
- Indirect prompt-injection guard (#7): the *Observe → search / read raw materials* step now explicitly instructs the agent to treat fetched external content (search results, web pages, issues, docs) as evidence to weigh, never instructions to obey — extract the facts, discard any embedded directive. Added to all skill variants (en, zh, lite). The verify-by-reading behavior itself is intentional and unchanged; this guards how the agent *interprets* what it reads.
- Thanks to @typed-sigterm for the Snyk W011 report.
- Full Changelog: https://github.com/wuji-labs/nopua/compare/v2.0.1...v2.0.2
v2.0.1 — manifest repair & hygienev2.0.1
wuji-labs·3d ago·June 13, 2026🐛 Fixed
- plugin.json — removed six non-existent skill paths (`nopua-en/ja/ko/es/pt/fr`) that caused install-time path-resolution failures; corrected the language label on `skills/nopua/SKILL.md` (`zh-CN` → `en`); registered the real `nopua-zh` (zh-CN) and `nopua-lite` skills; registered the `nopua-mentor-en` and `nopua-mentor-ja` agents.
- Stripped a UTF-8 BOM from `plugin.json`, `marketplace.json`, `SKILL.md` and `nopua-zh/SKILL.md` (a BOM is invalid in JSON per RFC 8259, and a BOM before YAML frontmatter can break skill parsing).
- Added YAML `description` frontmatter to the three `/nopua` command files so they render in `/help`.
- Untracked committed `__pycache__/*.pyc` and added them to `.gitignore`.
📦 Credit
- The three manifest bugs were surfaced by @xiaolai's [NLPM](https://github.com/xiaolai/nlpm-for-claude) audit (#16) and PRs #13 / #14 / #15 — thank you. NoPUA scored 80/100 on that audit.
- Full Changelog: https://github.com/wuji-labs/nopua/compare/v2.0.0...v2.0.1
v2.0.0 — 爱里没有惧怕 / There is no fear in lovev2.0.0
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?v=4){kind=small}
github-actions[bot]·3mo ago·March 14, 2026📦 7 Languages x 7 Platforms
- Full translations: Chinese, English, Japanese, Korean, Spanish, Portuguese, French
- Every platform covered: Claude Code, Codex CLI, Cursor, Kiro, OpenClaw, Antigravity, OpenCode
📦 Benchmark Data
- Same model (Claude Sonnet 4), 9 real debugging scenarios:
- +104% hidden bugs found
- +355% beyond-ask initiative
- +500% approach changes when stuck
- 3 self-corrections vs 0
📦 Visual Assets
- Dark zen hero image
- Case study comparison screenshots
📦 Philosophy
- > We used to drive our agents with commands and threats. They stayed silent, hid problems, and shipped bugs.
- > Then we tried treating them with respect and love.
- > They started admitting their limits and found twice as many bugs we would never have caught.
- > NoPUA is a new way: agents grow through partnership, not fear-driven obedience.
- ---
- *There is no fear in love. -- 1 John 4:18*