add-thinker

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/add-thinker — Codify a Thinker's Framework as a Skill

/add-thinker — 将思想家的思维框架转化为技能

Takes a thinker (person, book, school of thought) and produces a new Claude Code skill that applies their thinking to business ideas — the same way /munger applies Charlie Munger's mental lattice.

该技能接收一位思想家（人物、著作、思想流派），生成一个全新的Claude Code技能，将其思维方式应用于商业创意——就像/munger应用Charlie Munger的思维格栅一样。

What This Skill Does

该技能的功能

Deep research the thinker's framework — primary sources, talks, books, interviews
Extract the generic form — the reusable mental models, not just anecdotes
Design specialist agents — each covering a distinct lens from the framework
Synthesize a SKILL.md — a fully functional skill that spawns a team and produces a structured analysis, verdict, and actionable output

The output is a new skill file at

.claude/skills/<thinker>/SKILL.md

in the current repo, immediately usable as

/<thinker>

深度研究思想家的思维框架——包括原始资料、演讲、著作、访谈
提炼通用形式——提取可复用的思维模型，而非仅仅是轶事
设计专业Agent——每个Agent覆盖框架中的一个独特视角
合成SKILL.md文件——一个功能完整的技能，可生成一个分析团队，产出结构化的分析报告、结论和可执行的输出结果

输出的新技能文件位于当前仓库的

.claude/skills/<thinker>/SKILL.md

路径下，可立即通过

/<thinker>

调用。

Invocation

调用方式

/add-thinker <prompt>

The prompt can be:

A person:
```
Charlie Munger
```
,
```
Andy Grove
```
,
```
Nassim Taleb
```

A person + work:

Andy Grove — Only the Paranoid Survive

A school of thought:

Toyota Production System / Lean Thinking

A concept:
```
Nassim Taleb's Antifragility framework
```

A vague request:

that Intel CEO who wrote about strategic inflection points

If the prompt is too vague to identify a thinker, ask ONE clarifying question.

/add-thinker <prompt>

提示内容可以是：

人物：
```
Charlie Munger
```
、
```
Andy Grove
```
、
```
Nassim Taleb
```

人物+著作：

Andy Grove — Only the Paranoid Survive

思想流派：

Toyota Production System / Lean Thinking

概念：
```
Nassim Taleb's Antifragility framework
```

模糊请求：

那位撰写战略转折点相关内容的英特尔CEO

如果提示内容过于模糊无法识别思想家，只需提出一个澄清问题。

Phase 1: Identify and Scope

第一阶段：识别与界定

Parse the prompt to determine:

The thinker: Name, era, domain
The core works: Books, talks, essays that contain the framework
The domain: Business strategy, psychology, investing, engineering, etc.
The slug: lowercase hyphenated name for the skill directory (e.g.,
```
grove
```
,
```
taleb
```
,
```
toyota
```
)

Present back to the user:

undefined

解析提示内容以确定：

思想家：姓名、所处时代、研究领域
核心著作：包含该思维框架的书籍、演讲、文章
领域：商业战略、心理学、投资、工程学等
简称（slug）：技能目录使用的小写连字符命名（例如：
```
grove
```
、
```
taleb
```
、
```
toyota
```
）

向用户反馈如下内容：

undefined

Adding Thinker: [Name]

添加思想家：[姓名]

Core works to research:

[Book/talk 1]
[Book/talk 2]
[Book/talk 3]

Domain: [strategy / investing / psychology / engineering / etc.] Skill name: /[slug]

I'll now deep-research this framework, extract the generic mental models, and synthesize a skill. This takes a few minutes.

Starting research...

undefined

待研究的核心著作：

[书籍/演讲1]
[书籍/演讲2]
[书籍/演讲3]

领域： [战略 / 投资 / 心理学 / 工程学 / 等] 技能名称： /[简称]

我将深入研究该框架，提取通用思维模型，并合成技能。此过程需要几分钟时间。

开始研究...

undefined

Phase 2: Deep Research (Parallel Agents)

第二阶段：深度研究（并行Agent）

Spawn 3-4 research agents in parallel. Each focuses on a different aspect of the thinker's framework. Use

model: "sonnet"

for researchers.

并行生成3-4个研究Agent，每个Agent专注于思想家框架的不同方面。研究人员使用

model: "sonnet"

模型。

Agent 1: Primary Source Researcher

Agent 1：原始资料研究员

You are researching [THINKER]'s core framework for the purpose of creating
a reusable analytical tool.

Use WebSearch and WebFetch to find:

1. PRIMARY SOURCES
   - Full transcripts or detailed summaries of their key talks/speeches
   - Book summaries with actual frameworks extracted (not just reviews)
   - Interviews where they explain their thinking process
   - Check specific known sources: [list known URLs if any — Stripe Press,
     Farnam Street, personal websites, university lectures]

2. THE CORE FRAMEWORK
   - What are the 3-7 key principles or mental models?
   - How do they structure their analysis of a problem?
   - What questions do they always ask?
   - What is their equivalent of Munger's "inversion" or "lollapalooza"?
   - What is their unique contribution — the thing only THEY see?

3. THEIR VOCABULARY
   - Key terms they coined or use distinctively
   - Metaphors and analogies they rely on
   - Their catchphrases and memorable formulations

Report back with detailed findings including specific quotes and source URLs.
Be thorough — this research becomes the foundation of a permanent skill.

你正在研究[THINKER]的核心框架，目的是创建一个可复用的分析工具。

使用WebSearch和WebFetch查找：

1. 原始资料
   - 关键演讲/讲话的完整文字记录或详细摘要
   - 提取了实际框架的著作摘要（而非仅书评）
   - 思想家解释其思维过程的访谈
   - 检查特定已知来源：[列出已知URL（如有）——Stripe Press、Farnam Street、个人网站、大学讲座]

2. 核心框架
   - 有哪些3-7个关键原则或思维模型？
   - 他们如何构建问题分析的结构？
   - 他们总会提出哪些问题？
   - 他们与Munger的“逆向思维”或“lollapalooza效应”等价的概念是什么？
   - 他们的独特贡献是什么——只有他们能洞察到的点？

3. 专属词汇
   - 他们创造或独特使用的关键术语
   - 他们常用的隐喻和类比
   - 他们的口头禅和令人印象深刻的表述

返回包含具体引用和来源URL的详细研究结果。请务必详尽——此研究将成为永久技能的基础。

Agent 2: Applied Examples Researcher

Agent 2：应用案例研究员

You are researching how [THINKER] applies their framework to real-world cases.

Use WebSearch and WebFetch to find:

1. CASE STUDIES
   - Specific businesses, decisions, or situations they analyzed
   - How they walked through their framework step by step
   - What conclusions they reached and why
   - Cases where their framework predicted correctly
   - Cases where it failed or had blind spots

2. THE GENERIC PATTERN
   - Across all their case studies, what's the repeated analytical move?
   - What do they always check first?
   - What do they always check last?
   - What's their equivalent of "does the math work" or "what kills this"?

3. COMPARISON TO OTHER THINKERS
   - How does their framework overlap with Munger's lattice?
   - Where does it diverge or add something Munger misses?
   - What's complementary vs. contradictory?

Report with specific examples, quotes, and sources.

你正在研究[THINKER]如何将其框架应用于现实案例。

使用WebSearch和WebFetch查找：

1. 案例研究
   - 他们分析过的具体企业、决策或场景
   - 他们如何逐步应用框架进行分析
   - 他们得出了哪些结论及原因
   - 框架预测正确的案例
   - 框架失效或存在盲点的案例

2. 通用模式
   - 在所有案例研究中，反复出现的分析步骤是什么？
   - 他们总是首先检查什么？
   - 他们总是最后检查什么？
   - 他们与“数据是否合理”或“什么会导致失败”等价的检查项是什么？

3. 与其他思想家的对比
   - 他们的框架与Munger的思维格栅有哪些重叠？
   - 在哪些方面存在差异或补充了Munger未涵盖的内容？
   - 哪些内容是互补的，哪些是矛盾的？

返回包含具体案例、引用和来源的研究结果。

Agent 3: Counter-Arguments and Limitations Researcher

Agent 3：反对观点与局限性研究员

You are researching the limitations, critiques, and failure modes of
[THINKER]'s framework.

Use WebSearch and WebFetch to find:

1. KNOWN CRITIQUES
   - Academic or practitioner criticism of their framework
   - Cases where following their advice led to bad outcomes
   - Blind spots they acknowledge themselves
   - What types of problems does their framework NOT apply to?

2. FAILURE MODES
   - When does this thinking lead you astray?
   - What biases does the thinker themselves exhibit?
   - What does the framework miss that other frameworks catch?

3. CIRCLE OF COMPETENCE
   - What domains is this framework strongest in?
   - What domains should it NOT be applied to?
   - What's the thinker's own circle of competence vs. where they opine?

This is critical — every skill needs a "when NOT to use this" section.
Report with specific examples and honest assessment.

你正在研究[THINKER]框架的局限性、批评意见和失效模式。

使用WebSearch和WebFetch查找：

1. 已知批评
   - 学术界或从业者对其框架的批评
   - 遵循其建议导致不良结果的案例
   - 他们自己承认的盲点
   - 该框架不适用于哪些类型的问题？

2. 失效模式
   - 这种思维方式在什么情况下会误导你？
   - 思想家自身存在哪些偏见？
   - 该框架遗漏了哪些其他框架能捕捉到的内容？

3. 能力圈
   - 该框架在哪些领域表现最强？
   - 该框架不应应用于哪些领域？
   - 思想家自身的能力圈与他们发表意见的领域有何区别？

这一点至关重要——每个技能都需要包含“何时不应使用本技能”的部分。请返回包含具体案例和客观评估的研究结果。

Agent 4: Adjacent Thinkers and Synthesis (optional, spawn if the framework is broad)

Agent 4：相关思想家与综合研究（可选，若框架范围较广则生成）

You are researching thinkers adjacent to [THINKER] who extend, complement,
or challenge their framework.

Use WebSearch and WebFetch to find:

1. INTELLECTUAL LINEAGE
   - Who influenced this thinker?
   - Who did this thinker influence?
   - What's the "school of thought" this belongs to?

2. COMPLEMENTARY FRAMEWORKS
   - Other thinkers whose models stack well with this one
   - Specific models from other disciplines that strengthen this framework
   - What would a "lattice" look like that includes this thinker?

3. SYNTHESIS OPPORTUNITIES
   - How could this framework be combined with Munger's lattice?
   - What does this thinker add to the /munger analysis that's missing?
   - Could this be an "add-on module" to /munger rather than standalone?

Report with specific frameworks and how they interconnect.

你正在研究与[THINKER]相关的思想家，他们的观点对该框架起到延伸、补充或挑战作用。

使用WebSearch和WebFetch查找：

1. 学术传承
   - 谁影响了这位思想家？
   - 这位思想家影响了谁？
   - 这属于哪个“思想流派”？

2. 互补框架
   - 哪些其他思想家的模型能与该框架很好地结合？
   - 其他学科中哪些特定模型能强化该框架？
   - 包含该思想家的“思维格栅”会是什么样子？

3. 综合机会
   - 该框架如何与Munger的思维格栅结合？
   - 该思想家为/munger分析补充了哪些缺失的内容？
   - 它是否可以作为/munger的“附加模块”而非独立技能？

返回包含具体框架及其相互关联方式的研究结果。

Phase 3: Extract the Generic Form

第三阶段：提炼通用形式

After all research agents report back, the lead synthesizes the findings into a structured framework. This is the most important step — it's where raw research becomes a reusable analytical tool.

所有研究Agent返回结果后，主导Agent将研究结果整合为结构化框架。这是最重要的一步——将原始研究转化为可复用的分析工具。

The Extraction Template

提炼模板

For each thinker, extract:

1. THE CORE QUESTION
   What single question does this thinker's framework answer?
   - Munger: "Is this a good business to own for decades?"
   - Grove: "Are we at a strategic inflection point?"
   - Taleb: "Is this fragile, robust, or antifragile?"

2. THE KEY PRINCIPLES (3-7)
   The reusable mental models, stated as actionable rules.
   Each principle needs:
   - Name (their term or a clear label)
   - One-sentence rule
   - The mechanism (why it works)
   - How to apply it (specific questions to ask)
   - Example from the thinker's own work

3. THE ANALYTICAL PROCESS
   The step-by-step sequence they follow:
   - What do they check first? (the "no-brainer" equivalent)
   - What math do they run? (the numerical check)
   - What do they invert? (the "how does this die" check)
   - What's their synthesis move? (the "lollapalooza" equivalent)
   - What's their verdict framework? (the "In/Out/Too Tough" equivalent)

4. THE SPECIALIST LENSES
   Map to 3-5 agent roles, each covering a distinct analytical lens:
   - What discipline does each lens draw from?
   - What specific questions does each lens ask?
   - What output format does each lens produce?
   - How do the lenses interact? (cross-references)

5. THE VERDICT FRAMEWORK
   How does this thinker make a final call?
   - What are their "baskets" (equivalent to In/Out/Too Tough)?
   - What evidence tips the verdict?
   - What's their signature voice/style for delivering it?

6. THE FAILURE MODES
   When should you NOT use this framework?
   - Domain limitations
   - Known blind spots
   - Types of problems it misleads on

7. THE VOICE
   How does this thinker communicate?
   - Direct/indirect? Technical/colloquial? Serious/humorous?
   - Signature phrases, metaphors, rhetorical moves
   - What would they actually SAY about your idea?

Present this extraction to the user:

undefined

针对每位思想家，提取以下内容：

1. 核心问题
   该思想家的框架旨在回答哪个核心问题？
   - Munger：“这是一个值得持有数十年的好生意吗？”
   - Grove：“我们是否处于战略转折点？”
   - Taleb：“这是脆弱的、稳健的还是反脆弱的？”

2. 关键原则（3-7个）
   可复用的思维模型，以可执行规则的形式呈现。
   每个原则需包含：
   - 名称（他们使用的术语或清晰的标签）
   - 一句话规则
   - 作用机制（为何有效）
   - 应用方法（需提出的具体问题）
   - 思想家自身作品中的案例

3. 分析流程
   他们遵循的分步流程：
   - 首先检查什么？（相当于“无需思考的基础项”）
   - 进行哪些数据验证？（数值检查）
   - 进行哪些逆向思考？（相当于“如何导致失败”的检查）
   - 他们的综合步骤是什么？（相当于“lollapalooza效应”的步骤）
   - 他们的结论框架是什么？（相当于“接受/拒绝/难以判断”的框架）

4. 专业视角
   对应3-5个Agent角色，每个角色覆盖一个独特的分析视角：
   - 每个视角源自哪个学科？
   - 每个视角提出哪些具体问题？
   - 每个视角生成何种输出格式？
   - 视角之间如何互动？（交叉参考）

5. 结论框架
   这位思想家如何做出最终判断？
   - 他们的“分类”（相当于接受/拒绝/难以判断）是什么？
   - 哪些证据会影响结论？
   - 他们传达结论时的标志性风格是什么？

6. 失效模式
   何时不应使用该框架？
   - 领域局限性
   - 已知盲点
   - 会产生误导的问题类型

7. 语言风格
   这位思想家的沟通方式是怎样的？
   - 直接/间接？技术化/口语化？严肃/幽默？
   - 标志性短语、隐喻、修辞手法
   - 他们对你的创意实际会如何评价？

向用户展示提炼结果：

undefined

Framework Extraction: [Thinker]

框架提炼：[思想家]

Core question: [one sentence]

Key principles:

[Name] — [one-line rule]
[Name] — [one-line rule]
...

Specialist lenses (will become agents):

[Agent name] — [what they analyze]
[Agent name] — [what they analyze]
...

Verdict framework: [how the thinker makes a final call]

Voice: [how they communicate]

Not for: [when to NOT use this]

Does this capture the framework correctly? Anything to add or adjust?


Wait for user confirmation before generating the skill.

核心问题： [一句话]

关键原则：

[名称] — [一句话规则]
[名称] — [一句话规则]
...

专业视角（将转化为Agent）：

[Agent名称] — [分析内容]
[Agent名称] — [分析内容]
...

结论框架： [思想家做出最终判断的方式]

语言风格： [他们的沟通方式]

不适用场景： [何时不应使用本框架]

该提炼是否准确捕捉了框架内容？是否有需要添加或调整的地方？


在生成技能前等待用户确认。

Phase 4: Generate the Skill

第四阶段：生成技能

Using the extracted framework, generate a SKILL.md that follows the same architecture as /munger. The skill MUST include:

利用提炼的框架，生成遵循/munger架构的SKILL.md文件。技能必须包含以下部分：

Required Sections

必填章节

Frontmatter — name, description, allowed-tools (same set as /munger)
Header — skill name, one-paragraph description of what it does
Core Principles — the thinker's key principles, stated as non-negotiable rules for the analysis (equivalent to Munger's "five notions")
Invocation — how to trigger, what arguments to provide
Phase 1: Understand the Idea — lead gathers context, presents understanding
Phase 2: Spawn the Team — detailed prompts for each specialist agent. Each agent prompt MUST include:
- Role and discipline
- The business idea (substituted at runtime)
- Specific analytical questions from the framework
- Output format
- Cross-reference instructions for messaging teammates
- The thinker's actual vocabulary and framing
Phase 3: Monitor & Cross-Pollinate — same as /munger
Phase 4: Synthesize — The [Thinker] Verdict — the lead's synthesis process. MUST include:
- How to combine agent findings
- The framework's equivalent of "lollapalooza detection"
- The verdict framework (the thinker's version of In/Out/Too Tough)
- A "What [Thinker] Would Say" section written in their voice
- Actionable rules derived from the analysis
Phase 5: Present & Follow-up — summary, verdict, next steps
Batch Mode — how to compare multiple ideas
Scoring Discipline — honesty rules, evidence requirements
Important Notes — cost, model selection, pairing with other skills

前置信息——名称、描述、允许使用的工具（与/munger一致）
标题——技能名称、一段功能描述
核心原则——思想家的关键原则，作为分析的不可协商规则（相当于Munger的“五个理念”）
调用方式——触发方式、需提供的参数
第一阶段：理解创意——主导Agent收集背景信息，呈现理解结果
第二阶段：生成团队——每个专业Agent的详细提示。每个Agent提示必须包含：
- 角色与所属学科
- 商业创意（运行时替换）
- 框架中的具体分析问题
- 输出格式
- 与其他Agent沟通的交叉参考指令
- 思想家实际使用的词汇和表述方式
第三阶段：监控与交叉验证——与/munger一致
第四阶段：综合——[思想家]结论——主导Agent的综合流程。必须包含：
- 如何整合Agent的研究结果
- 框架中与“lollapalooza效应检测”等价的部分
- 结论框架（思想家版本的接受/拒绝/难以判断）
- “思想家会如何评价”章节，采用其语言风格撰写
- 从分析中得出的可执行规则
第五阶段：展示与跟进——摘要、结论、后续步骤
批量模式——如何对比多个创意
评分准则——诚实规则、证据要求
重要说明——成本、模型选择、与其他技能搭配使用

Quality Requirements

质量要求

Agent prompts must be LONG and SPECIFIC — not "analyze the economics" but detailed questions with the thinker's actual vocabulary and examples. Look at the /munger agent prompts for the standard. Each should be 30-50 lines.
The verdict must be HONEST — capture the thinker's actual standards. If they're a harsh critic (like Munger), the skill should reject most ideas. If they're an optimist, the skill should reflect that — but still have rigor.
The voice must be AUTHENTIC — the "What [Thinker] Would Say" section should sound like them, using their actual phrases and rhetorical style.
Cross-references to /munger — note where this framework overlaps with or complements Munger's lattice. Suggest pairing where appropriate.

Agent提示必须详尽且具体——不应是“分析经济情况”，而应是包含思想家实际词汇和案例的详细问题。参考/munger的Agent提示标准，每个提示应包含30-50行内容。
结论必须真实——体现思想家的实际标准。如果他们是严苛的批评者（如Munger），技能应拒绝大多数创意。如果他们是乐观主义者，技能也应反映这一点——但仍需保持严谨性。
语言风格必须真实——“思想家会如何评价”章节应听起来像他们本人，使用他们的实际短语和修辞风格。
与/munger的交叉参考——注明该框架与Munger思维格栅的重叠或互补之处。在合适的情况下建议搭配使用。

File Output

文件输出

Write the skill to:

.claude/skills/<slug>/SKILL.md

Also copy it to the global skills directory so it's available everywhere:

~/.claude/skills/<slug>/SKILL.md

将技能写入以下路径：

.claude/skills/<slug>/SKILL.md

同时复制到全局技能目录，使其可在所有环境中使用：

~/.claude/skills/<slug>/SKILL.md

Phase 5: Verify and Present

第五阶段：验证与展示

After writing the skill:

Read it back — verify it's syntactically correct and complete
Check it appears — the skill should show up in the skills list
Present to the user:

undefined

写入技能文件后：

回读文件——验证语法正确且内容完整
检查可用性——技能应出现在技能列表中
向用户展示：

undefined

New Thinker Skill: /<slug>

新思想家技能：/<slug>

Framework: [one-sentence description] Core question: [what it answers] Agents: [N] specialists

[Agent] — [lens]
[Agent] — [lens] ... Verdict: [framework's decision categories] Voice: [how it communicates]

Installed at:

.claude/skills/<slug>/SKILL.md (this repo)
~/.claude/skills/<slug>/SKILL.md (global)

Try it: /<slug> [your business idea]

Suggested workflow:

/garrytan — refine the idea
/munger — Munger's lattice
/<slug> — [thinker]'s framework

undefined

框架： [一句话描述] 核心问题： [解决的问题] Agent： [数量]个专业Agent

[Agent名称] — [视角]
[Agent名称] — [视角] ... 结论： [框架的决策分类] 语言风格： [沟通方式]

安装路径：

.claude/skills/<slug>/SKILL.md（当前仓库）
~/.claude/skills/<slug>/SKILL.md（全局）

试用：/<slug> [你的商业创意]

建议工作流：

/garrytan — 优化创意
/munger — Munger思维格栅分析
/<slug> — [思想家]框架分析

undefined

Architecture Notes

架构说明

Research agents use sonnet — they're doing web search and extraction, not deep reasoning. The lead (opus) handles synthesis and skill generation.
3-4 research agents max — more than that produces diminishing returns and the lead can't synthesize well beyond 4 perspectives.
The generated skill follows /munger's architecture exactly — same phase structure, same agent spawning pattern, same verdict format. This makes all thinker skills composable and familiar.
Each thinker skill is standalone — it doesn't depend on /munger being installed. But the output format is compatible, so you can run both and compare verdicts.
The global install means the skill persists — even if you delete this repo, the thinker skill remains available in
```
~/.claude/skills/
```
.

研究Agent使用sonnet模型——他们负责网页搜索和提取，而非深度推理。主导Agent（opus）负责综合和技能生成。
最多生成3-4个研究Agent——超过4个会产生边际收益递减，主导Agent也难以很好地综合4个以上的视角。
生成的技能严格遵循/munger的架构——相同的阶段结构、相同的Agent生成模式、相同的结论格式。这使得所有思想家技能可组合且易于使用。
每个思想家技能都是独立的——不依赖/munger的安装。但输出格式兼容，因此可以同时运行两者并对比结论。
全局安装确保技能持久存在——即使删除当前仓库，思想家技能仍会保留在
```
~/.claude/skills/
```
路径下。

Examples of Thinkers This Should Work For

适用的思想家示例

Prompt	Skill	Core Question
Andy Grove	/grove	Are we at a strategic inflection point?
Nassim Taleb antifragility	/taleb	Is this fragile or antifragile?
Peter Thiel Zero to One	/thiel	Is this a 0-to-1 or 1-to-n business?
Toyota Production System	/toyota	Where is the waste and how do we eliminate it?
Ben Thompson Stratechery	/thompson	What's the aggregation theory play here?
Clayton Christensen	/christensen	Is this disruptive or sustaining innovation?
Hamilton Helmer 7 Powers	/helmer	Which of the 7 powers does this business have?
Jeff Bezos	/bezos	Is this a one-way or two-way door decision?
Ray Dalio Principles	/dalio	What principles govern this situation?
Eliyahu Goldratt Theory of Constraints	/goldratt	What's the bottleneck?

提示内容	技能	核心问题
Andy Grove	/grove	我们是否处于战略转折点？
Nassim Taleb antifragility	/taleb	这是脆弱的还是反脆弱的？
Peter Thiel Zero to One	/thiel	这是0到1还是1到N的生意？
Toyota Production System	/toyota	浪费在哪里，如何消除？
Ben Thompson Stratechery	/thompson	这里的聚合理论应用点是什么？
Clayton Christensen	/christensen	这是颠覆性创新还是持续性创新？
Hamilton Helmer 7 Powers	/helmer	这项业务拥有7种力量中的哪一种？
Jeff Bezos	/bezos	这是单向门还是双向门决策？
Ray Dalio Principles	/dalio	哪些原则适用于此场景？
Eliyahu Goldratt Theory of Constraints	/goldratt	瓶颈是什么？

What This Skill Does NOT Do

该技能不具备的功能

It does not evaluate business ideas itself — it creates tools that do.
It does not replace reading the thinker's actual work — the research phase extracts the framework, but the skill description should reference primary sources so users can go deeper.
It does not guarantee the generated skill is perfect on first pass — complex thinkers may need iteration. The user can edit the SKILL.md after generation.

不直接评估商业创意——它创建用于评估的工具。
不能替代阅读思想家的原著——研究阶段提取框架，但技能描述应参考原始资料，以便用户深入学习。
不保证生成的技能一次就完美——复杂的思想家可能需要迭代优化。用户可以在生成后编辑SKILL.md文件。