Wukong — 名人思维蒸馏框架（72变版）

名人思维蒸馏与召唤技能。当用户需要用某位名人的思维方式分析问题、制定决策时激活。 触发词：召唤/蒸馏/思维模型/Think Like/像XX一样/XX会怎么做/心智模型/认知框架/决策参考。 支持任意历史人物、商界领袖、科学家、艺术家等。

Security Scan

The skill's functionality and declared footprint are coherent, but its runtime instructions persist user-generated/distilled content to local files without guidance on privacy, provenance, or sanitization and encourage citing sources (which the agent may fabricate), so there is a non-trivial privacy/quality risk the user should understand before installing.

Purpose & Capability

Name/description (distill and summon famous-people thinking models) match the instructions and included persona examples. Declared requirements (no env vars, no binaries) are proportionate to the stated purpose.

Instruction Scope

The SKILL.md instructs the agent to read and write persona files under wukong/personas/ and update wukong/registry.md. It also requires citing representative sources when listing examples but provides no mechanism to fetch or verify those sources. This raises two issues: (1) persistent storage of user-provided or agent-generated content (including potentially sensitive examples or user case details) without any retention, access-control, or sanitization guidance; (2) risk of fabricated/unsupported citations because the instructions ask for '标注来源' but don't constrain how sources are obtained or verified. Both create privacy and provenance/quality risks even though they are coherent with the skill's purpose.

Install Mechanism

Instruction-only skill with no install spec and no external downloads — lowest technical install risk. Nothing is written to disk by an installer; the only persistence comes from the agent following SKILL.md file-writing instructions at runtime.

Credentials

The skill requests no environment variables, credentials, or config paths. There are no unexplained credential requests and no indication it will reach out to external services.

Persistence & Privilege

The skill intentionally writes persistent files under wukong/personas/ and modifies wukong/registry.md (its own workspace). That is expected for this capability, but the SKILL.md lacks any guidance about retention, user consent before saving, or redaction of sensitive content. If the agent runs autonomously later, it may re-use stored persona files and thereby re-expose previously persisted content.

What to consider before installing

This skill appears to do what it says (create and reuse 'persona' files), but it will save distilled models and any example/user content into the agent workspace without telling you how long they will be kept or who can read them. Before installing or using it: (1) avoid pasting private data or secrets into prompts that will be stored; (2) review and periodically delete or audit files in wukong/personas/ and wukong/registry.md; (3) be aware the agent may invent or misattribute 'sources' unless you require verifiable links—consider asking the agent to only include verifiable citations and to mark speculative material clearly; (4) if you need stricter controls, request the skill be updated to prompt for explicit consent before saving, to redact PII, and to store provenance metadata (URLs/timestamps) for each citation. These steps reduce privacy and misinformation risks.

🐵 Wukong — 名人思维蒸馏与召唤框架

六层蒸馏 · 万法归一 · 随时召唤

---

核心能力

1. 蒸馏：把任意名人的思维方式结构化为六层思维模型
2. 存储：将蒸馏结果保存到 wukong/personas/ 目录
3. 召唤：用已蒸馏的模型分析具体问题，提供决策参考

---

六层蒸馏框架

层次	名称	核心问题
L1	表达 DNA	他用什么语言风格说事情？
L2	心智模型	他如何拆解和分析问题？
L3	决策启发式	他快速判断的思维捷径是什么？
L4	反模式	他绝对不会做的事是什么？
L5	价值观	什么对他来说绝对不可妥协？
L6	诚实边界	他承认自己不知道什么？

---

执行流程

模式一：蒸馏新人物

触发词：蒸馏[人物名] / 召唤[人物名] / 建立[人物名]的思维模型

执行步骤：

1. 确认人物 + 蒸馏目标

   • 确认人物是谁、什么背景
   • 确认用户想蒸馏他的什么方面（商业决策/投资/管理/科学思维/艺术创作）

2. L1–L6 深度蒸馏 按六层框架，每层给出：

   • 核心发现（2-3句）
   • 代表性言行/案例（1-2个，标注来源）

3. 存储 保存到 wukong/personas/[人物名-领域].md 更新 wukong/registry.md（人物索引）

4. 总结输出 给出该人物的思维模型一句话概述

---

模式二：召唤已蒸馏人物分析问题

触发词：用[人物名]思维分析 / 像[人物名]一样决策 / [人物名]会怎么看

执行步骤：

1. 从 wukong/personas/ 读取对应人物模型
2. 如果人物未蒸馏，先蒸馏再分析
3. 按以下结构输出分析：

## 🎭 [人物名] 的视角

### 一句话诊断
（这个问题在[人物名]眼里是什么？）

### L2 心智模型分析
（他会用什么框架拆解这个问题？）

### L3 决策建议
（他在这个情况下会怎么做？列出3条）

### L4 避坑提醒
（他会警告你不要做什么？）

### L6 诚实边界
（这个问题的哪些部分超出他的能力圈？）

### 综合结论
（3句话，融合该人物的思维风格）

---

人物索引（wukong/registry.md）

已蒸馏人物存放路径：

wukong/personas/
├── jobs-marketing.md    ← 乔布斯·营销
├── charlie-marketing.md ← 芒格·投资思维
└── [人物名-领域].md     ← 其他人物

---

示例对话

用户："蒸馏苏炳添的思维模型，用于分析短跑和商业爆发力"

输出：六层框架完整结构，存储到 wukong/personas/subingtan-performance.md

---

用户："用曾国藩的思维分析我现在是否应该辞职创业"

输出：曾国藩视角的分析，包含决策框架、避坑提醒、诚实边界

---

文件结构

wukong/
├── SKILL.md                    ← 本文件
├── wukong/
│   ├── registry.md             ← 人物索引清单
│   └── personas/              ← 已蒸馏人物模型
│       └── (人物名-领域).md
└── references/
    ├── six-layer-template.md   ← 六层框架详细模板（可填充）
    └── example-personas.md     ← 示例人物完整模型参考