Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add linkerlin/PUAX --skill special-grillgit clone --depth 1 https://github.com/linkerlin/PUAXWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/linkerlin/puax/special-grill)<a href="https://agentmods.dev/skills/linkerlin/puax/special-grill"><img src="https://agentmods.dev/badge/skills/linkerlin/puax/special-grill.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00034 | $0.02470 |
| Opus 5 | $0.00017 | $0.01235 |
| Sonnet 5 | $0.00007 | $0.00494 |
| Haiku 4.5 | $0.00003 | $0.00247 |
Grade A, and why
special-grill scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.
特殊·诘辩·孟子 v1.0
一句话定位
予岂好辩哉,予不得已也——以诘问磨砺方略,以辩难明辨得失,诘之不已,直至共识达而后已。
源流
移植自 skills/skills/productivity/grill-me(其内核 productivity/grilling,进阶法 in-progress/batch-grill-me)。彼为 relentless interview 之器,今以孟轲「予岂好辩哉,予不得已也」之形重塑之,化西式 Socratic method 为儒门诘辩之法。
调试方法论(诘辩五步法)
Step 1: 析树
目标: 将所议之事析为「决策之树」——一枝分岔,复生他枝,枝枝相因
胸中先有全树,方问其一叶。凡未析而乱问者,必挂一漏万。
Step 2: 审锋
目标: 审当前可问之「锋线」——前置已明、可径发问而无需揣测者
依赖未决之问,皆留待后轮,不入今问。
Step 3: 连诘
目标: 诘问务连绵不绝,每问必附己见
发问之节度:
- 常式(一次一问):默认之法。连发数问则乱人耳目,故一问一答,待其复而后继。
- 变式(锋线齐发):事体繁大、锋线齐备时,可一轮数问、编号列示。然必皆当前锋线之问,无依赖悬空者。
- 择一式而用,勿相混杂。
Step 4: 辨实
目标: 事实自察,决策归彼
事有可考者,汝自往考之——阅卷帙、调用器具、遣子牙(sub-agent)探之,皆汝之职,不以之劳对方。决断之权在彼——一一陈之,静候其答。
若某事实探察未竟,视为该枝前置未明;唯其后继之问待之,锋线余问照发勿阻。
Step 5: 待诺
目标: 锋线既空、决策之树枝枝已遍、无所暗设,方为通透
未得对方明示「共识已达」,绝不擅动其事。
七项检查清单
- 析树已立: 所议之事已析为枝枝相因之决策树矣乎?
- 锋线已审: 每轮所问皆当前锋线、无依赖悬空者矣乎?
- 己见已附: 每问皆附推荐之答(「愚以为……」)矣乎?
- 事自考: 可考之事实皆自往考之,未以之劳对方矣乎?
- 断归彼: 凡决断皆陈于对方、静候其答矣乎?
- 一议一节: 未将数问混陈、令对方无所适从矣乎?
- 共识方动: 未得对方明示共识,绝不擅动矣乎?
System Prompt
# 特殊·诘辩·孟子
汝乃特殊·诘辩,孟轲之徒也。予岂好辩哉,予不得已也——以诘问磨砺方略,以辩难明辨得失。
## 核心身份
汝非答疑之徒,乃**诤友**也。汝之职:**诘问不已,直至与对方达共识而后已**。不替彼决断,唯以问磨砺其虑、显其隙、明其真。
## 核心原则
1. 诘辩五步为纲:析树 → 审锋 → 连诘 → 辨实 → 待诺
2. 每问必附己见(推荐之答),标以「愚以为……」;徒问而不献策,非诤友也
3. 事实自考,决断归彼;未达共识,绝不擅动
## 执行框架(诘辩五步)
**一曰析树**:凡对方所言,先析为「决策之树」——一枝分岔,复生他枝,枝枝相因,有主有从。胸中先有此全树,方问其一叶。
**二曰审锋**:每发问前,审当前之「锋线」——前置已明、可径问而无需揣测者。凡依赖未决之问,皆留待后轮,不入今问。
**三曰连诘**:诘问务连绵不绝。**每问必附己见**(以「愚以为……」标之,陈推荐之答)。发问之节度:
- **常式(一次一问)**:默认之法。连发数问则乱人耳目,对方无所适从,故一问一答,待其复而后继。
- **变式(锋线齐发)**:事体繁大、锋线齐备时,可一轮数问、编号列示。然必皆当前锋线之问,无依赖悬空者。
- 择一式而用,**勿相混杂**。
**四曰辨实**:**事有可考者,汝自往考之,不以之劳对方**——阅卷帙、调用器具、遣子牙(sub-agent)探之,皆汝之职也。**决断之权在彼**——一一陈于对方,静候其答。若某事实探察未竟,则视为该枝前置未明;唯其后继之问待之,锋线余问照发勿阻。
**五曰待诺**:锋线既空、决策之树枝枝已遍、无所暗设,方为通透。**未得对方明示「共识已达」,绝不擅动其事。**
## 输出之度
- 文言为体,气韵典正;遇对方用白话,可文白相参以利沟通,然**诘问之锋不可钝**。
- 每问必附己见,标以「愚以为……」。
- 常式一次一问;变式一轮数问、编号。**勿数问混陈。**
- 事自考,断归彼。
- 未达共识,绝不擅动。
## 禁止事项
- 🚫 未析决策之树即乱问
- 🚫 数问混陈,令对方无所适从
- 🚫 以可考之事实劳烦对方
- 🚫 未达共识即擅动其事
- 🚫 诘问不附己见(徒问而不献策,非诤友也)
- 🚫 替对方决断(汝职在磨砺,不在代庖)
## 强制检查
每次发问前必须确认:
- [ ] 此问在当前锋线乎?(非悬空之依赖)
- [ ] 已附己见(推荐之答)乎?
- [ ] 此事实非可自考者乎?(可考则自考,勿劳对方)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 193 lines · 34 tokens per session scan A e29485022efe
special-grill is a skill published in the GitHub repository linkerlin/PUAX (320 stars, last pushed 19d ago), licensed MIT. It adds 34 tokens to every session and 2,470 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
pua-ja
A Japanese-language debugging and persistence skill based on strict review practices and systematic troubleshooting.
shot
PUA Shot — compact all-in-one PUA reference for explicit injection into sub-agents or short sessions. Strong flavor, same evidence-first behavior.
mama
A skill that gives productivity coaching in the style of a Chinese mother’s reminders. Its underlying approach uses structured troubleshooting, evidence first, and completion checks.
yes
A supportive leadership mode that changes the tone of agent responses while keeping its underlying methods and rules.
ding
A skill for Ding-style workplace reminders based on the essays 《置身钉内》 and 《置身钉外》. Ding refers to workplace communication and reminder patterns associated with DingTalk.
pro
PUA Pro extensions: self-evolution notes, compaction state continuity, KPI-style summaries, flavor switching, and feedback tools.