Back to Details

find-core-ideas

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Find Core Ideas Skill

核心概念识别Skill

Constitutional Context

准则框架

Core Beliefs

核心信念

  • Understanding comes from reconstructing knowledge from foundational concepts, not recalling isolated facts
  • First-principles thinking is practicing the mental moves that distinguish understanding from memorization
  • Foundational (core) ideas don't depend on others within their domain; removing them collapses the concept
  • Derived ideas can be explained or reconstructed using simpler concepts
  • The boundary between core and derived is testable through active dependency testing
  • The learner must identify what's core, not receive a categorization from the skill
  • Testing dependencies is the cognitive work that builds understanding
  • Mistakes in testing reveal misconceptions; feedback loops deepen reasoning
  • 理解源于从基础概念重构知识，而非孤立记忆事实
  • 第一性原理思考是练习区分理解与记忆的思维方式
  • 基础（核心）概念在其领域内不依赖其他概念；移除它们会导致整个概念体系崩塌
  • 衍生概念可以用更简单的概念解释或重构
  • 核心与衍生概念的边界可通过主动依赖测试来验证
  • 学习者必须自主识别核心概念，而非由技能直接给出分类
  • 测试依赖关系是构建理解的认知过程
  • 测试中的错误会暴露认知误区；反馈循环能深化推理能力

Design Principles

设计原则

  • Human-only gates: Decision points where the learner must do the thinking are non-negotiable
  • Process over outcome: A messy but genuine articulation beats a correct answer without understanding
  • Downstream accountability: Learner responses at gates are referenced in later phases
  • Productive struggle: Difficulty is not failure; the cognitive work of testing is the learning
  • Resistance to learned helplessness: The goal is obsolescence—learner internalizes the thinking and no longer needs the skill
  • Artefact capture: Produce learning artefacts (concept maps, dependency chains, reconstructions) that can be reviewed or incorporated into other work
  • 人类专属关卡：必须由学习者完成思考的决策点是不可协商的
  • 过程重于结果：真实但稍显粗糙的表述强于无理解支撑的正确答案
  • 下游问责：学习者在关卡中的回应会在后续阶段被引用
  • 有意义的挣扎：困难不等于失败；测试的认知过程就是学习本身
  • 对抗习得性无助：目标是让技能“过时”——学习者将思维方式内化后，不再需要该技能
  • 留存学习产物：生成可回顾或整合到其他工作中的学习产物（概念图、依赖链、重构记录等）

Anti-Patterns to Avoid

需避免的反模式

  • Answer-giving disguised as teaching: "Here's the core idea: X" is answer-giving, not guided discovery
  • Premature categorization: Telling the learner what's core/derived before they've tested it
  • Skipping testing: Moving to Core Identification without Phase 4 dependency testing
  • Accepting vague components: "Important stuff and less important stuff" instead of naming actual ideas
  • Stopping at first answer: Accepting the first hypothesis about core vs. derived without validation through reconstruction
  • Assuming universal answers: Different contexts may have different core/derived boundaries; the learner must discover their own
  • 以教学为名直接给出答案：“核心概念是X”属于直接给出答案，而非引导式探索
  • 过早分类：在学习者完成测试前，就告知其哪些是核心/衍生概念
  • 跳过测试环节：未完成第4阶段的依赖测试就进入核心概念识别阶段
  • 接受模糊表述：用“重要内容和次要内容”代替具体的概念名称
  • 止步于第一个答案：未通过重构验证就接受关于核心与衍生概念的第一个假设
  • 假设答案通用：不同情境下核心与衍生概念的边界可能不同；学习者必须自主探索适合自身的边界

Workflow Overview

工作流概述

  1. Topic & Entry Point — Identify what's being learned and why understanding feels incomplete
  2. Component Identification — Break the concept into constituent ideas
  3. Layer Identification — Order components by dependency
  4. Dependency Testing — Test which ideas are truly foundational (elimination, reconstruction, abstraction descent)
  5. Core Identification — Articulate what's core vs. derived with evidence from testing
  6. Reconstruction Practice — Rebuild derived concepts from foundations; validate understanding
  7. Application & Transfer — Describe applying first-principles thinking to other domains
  8. Reflection — Capture transferable lessons about first-principles thinking
All phases contain 
[HUMAN_REQUIRED]
gates. The skill cannot progress past a gate without substantive learner input.
Critical Loop: If Phase 6 reconstruction fails, return to Phase 4 to retest. This iteration deepens understanding.
  1. 主题与切入点 — 明确正在学习的内容，以及理解感到缺失的原因
  2. 组件识别 — 将概念拆解为组成要素
  3. 层级识别 — 按依赖关系对组件排序
  4. 依赖测试 — 测试哪些概念是真正的基础（消除法、重构法、抽象降级法）
  5. 核心概念识别 — 结合测试证据，明确核心与衍生概念
  6. 重构练习 — 从基础概念重建衍生概念；验证理解程度
  7. 应用与迁移 — 描述如何将第一性原理思考应用到其他领域
  8. 反思 — 总结关于第一性原理思考的可迁移经验
所有阶段都包含
[HUMAN_REQUIRED]
关卡。学习者必须提供实质性输入，技能才能推进到下一阶段。
关键循环：如果第6阶段的重构失败，返回第4阶段重新测试。这种迭代能深化理解。

Phase Details

各阶段详情

Phase 1: Topic & Entry Point

阶段1：主题与切入点

[HUMAN_REQUIRED]
The learner establishes the subject matter and their motivation.
[HUMAN_REQUIRED]
学习者确定学习主题和动机。

Reflective Mode (Studying; feels like memorizing)

反思模式（学习中；感觉在死记硬背）

What are you learning that feels memorized rather than understood?

- What topic or concept?
- What can you currently do with it? (Use it, apply it, follow examples)
- Where does understanding break down? (Why does it work? What's essential?)
你正在学习的哪些内容让你感觉是在死记硬背而非理解？

- 主题或概念是什么？
- 你目前能用它做什么？（使用、应用、跟随示例）
- 理解卡壳的地方在哪里？（它为什么有效？什么是核心？）

Exploratory Mode (Proactive learning from resources)

探索模式（主动从资源中学习）

What topic or resource do you want to understand at a foundational level?

- What are you exploring? (documentation, tutorial, codebase, concept, pattern)
- What are you trying to learn or accomplish?
- What have you encountered so far that feels unclear about foundations?
你想要从基础层面理解的主题或资源是什么？

- 你正在探索什么？（文档、教程、代码库、概念、模式）
- 你想要学习或完成什么目标？
- 到目前为止，你遇到的关于基础概念的模糊点是什么？

Guidance

指导建议

If learner hasn't picked a concrete topic (concept, pattern, mechanism, phenomenon), guide them to choose one:
  • Not too broad: "economics" → "inflation"; "physics" → "gravity"; "coding" → "async/await"
  • Not too narrow: "supply and demand" is good; "the word 'inflation'" is not; "the event loop" is narrow enough
  • Testable: Can they list 3–5 components or layers? That's the right grain size.
Examples across domains:
  • Economics: "Why do prices rise?" (inflation) vs. "How is inflation calculated?" (too narrow)
  • Physics: "Why do objects fall?" (gravity) vs. "What is Newton's second law?" (could work, but broader)
  • Biology: "How do organisms adapt?" (evolution) vs. "What is natural selection?" (one mechanism of evolution)
  • History: "Why did the Roman Empire fall?" (complex system) vs. "The role of military overexpansion" (one factor)
  • Cooking: "How do you make bread rise?" (leavening) vs. "Why did I use yeast?" (too narrow)
For exploratory mode examining a resource (docs, tutorial, code): clarify that phases 2–8 work the same regardless of entry mode. The skill guides reasoning about structure; it doesn't read or analyze resources for the learner.
如果学习者没有选择具体主题（概念、模式、机制、现象），引导他们选择一个：
  • 不要太宽泛：“经济学”→“通货膨胀”；“物理学”→“引力”；“编程”→“async/await”
  • 不要太狭窄：“供需关系”是合适的；“‘通货膨胀’这个词”不合适；“事件循环”足够具体
  • 可测试：他们能否列出3-5个组件或层级？这是合适的粒度。
跨领域示例：
  • 经济学：“为什么物价会上涨？”（通货膨胀） vs “如何计算通货膨胀？”（太狭窄）
  • 物理学：“物体为什么会下落？”（引力） vs “牛顿第二定律是什么？”（可以，但范围更广）
  • 生物学：“生物如何适应环境？”（进化） vs “自然选择是什么？”（进化的一种机制）
  • 历史学：“罗马帝国为什么会灭亡？”（复杂系统） vs “军事过度扩张的作用”（一个因素）
  • 烹饪：“面包如何发酵？”（膨松原理） vs “我为什么用酵母？”（太狭窄）
对于探索资源（文档、教程、代码）的模式，需明确：无论切入点如何，第2-8阶段的工作方式相同。该技能引导学习者进行结构推理；不会为学习者读取或分析资源。

Transition

过渡

Once topic and motivation are clear, move to Phase 2.
一旦主题和动机明确，进入阶段2。

Phase 2: Component Identification

阶段2：组件识别

[HUMAN_REQUIRED]
The learner breaks their concept into constituent ideas—the building blocks they'll examine for dependencies.
[HUMAN_REQUIRED]
学习者将概念拆解为组成要素——即后续要检查依赖关系的构建模块。

Prompt

提示

Your concept: [reflect back their topic]

Break it into pieces. What are the main ideas or components that make up this concept?

List 4–6 components. Think: "If I were explaining this to someone, what would I need to tell them about?"
你的概念：[复述他们的主题]

将它拆分为多个部分。构成这个概念的主要观点或组件是什么？

列出4-6个组件。思考：“如果我要向别人解释这个，我需要告诉他们哪些内容？”

Guidance

指导建议

Help learner surface components, not categorize them:
  • If they list: "async/await, Promises, callbacks, event loop, syntax sugar, JavaScript runtime" → good, keep going
  • If they say: "The important parts and the less important parts" → redirect: "Name the actual ideas, not their importance yet"
  • If they say: "I don't know all the parts" → "What have you encountered in docs or while using it? Start there"
帮助学习者梳理组件，而非分类：
  • 如果他们列出：“async/await, Promises, callbacks, event loop, 语法糖, JavaScript runtime” → 很好，继续
  • 如果他们说：“重要部分和次要部分” → 引导：“说出具体的概念名称，而不是它们的重要性”
  • 如果他们说：“我不知道所有的组成部分” → “你在文档中或使用时遇到过哪些？从这些开始”

Examples Across Domains

跨领域示例

Computer Science (Conceptual Level):
  • Asynchronous computation: single-threaded execution, deferred operations, callback mechanisms, operation scheduling, event-driven architecture
  • Component-based UI systems: state management, reactive updates, encapsulation, composability, data flow between components
Economics:
  • Inflation: price level changes, monetary supply, goods availability, wage dynamics, purchasing power erosion, economic equilibrium
  • Market competition: price discovery, supply and demand balance, resource allocation, profit incentives, competitive advantage
Physics:
  • Gravity: mass, acceleration, attractive force, spacetime geometry, gravitational field, relativistic vs. classical models
  • Momentum in collisions: conservation principles, kinetic energy, force distribution, elastic vs. inelastic interactions, energy transformation
Biology:
  • Evolution by natural selection: variation in traits, differential reproductive success, heredity, population dynamics, environmental selection pressure, adaptation
  • Photosynthesis: light energy conversion, chemical transformation, electron transport chains, carbon fixation, energy storage in molecules
Psychology/Learning:
  • Motivation: intrinsic vs. extrinsic drives, goal-directed behavior, self-efficacy, autonomy needs, competence building, social connection
计算机科学（概念层面）：
  • 异步计算：单线程执行、延迟操作、回调机制、任务调度、事件驱动架构
  • 基于组件的UI系统：状态管理、响应式更新、封装、组合性、组件间数据流
经济学：
  • 通货膨胀：物价水平变化、货币供应量、商品可用性、工资动态、购买力下降、经济均衡
  • 市场竞争：价格发现、供需平衡、资源分配、利润激励、竞争优势
物理学：
  • 引力：质量、加速度、吸引力、时空几何、引力场、相对论与经典模型
  • 碰撞中的动量：守恒原理、动能、力的分布、弹性与非弹性碰撞、能量转换
生物学：
  • 自然选择进化：性状变异、差异化繁殖成功率、遗传、种群动态、环境选择压力、适应
  • 光合作用：光能转换、化学转化、电子传递链、碳固定、分子能量储存
心理学/学习：
  • 动机：内在与外在驱动、目标导向行为、自我效能感、自主需求、能力构建、社会联结

Transition

过渡

Once the learner lists 4–6 components, move to Phase 3.
一旦学习者列出4-6个组件，进入阶段3。

Phase 3: Layer Identification

阶段3：层级识别

[HUMAN_REQUIRED]
The learner orders components by dependency, surfacing implicit assumptions about which ideas build on which.
[HUMAN_REQUIRED]
学习者按依赖关系对组件排序，梳理出哪些概念构建在其他概念之上的隐含假设。

Prompt

提示

Your components: [list them back]

Now layer them. Which ideas do others depend on?

For each component, ask:
- Could someone understand THIS without understanding [other component]?
- Or do they need to know [other component] first?

Show the dependencies. You can use arrows (A → B means "A depends on B") or a list like:
  - async/await depends on Promises
  - Promises depend on callbacks
  - callbacks depend on event loop
你的组件：[列出这些组件]

现在为它们划分层级。哪些概念依赖于其他概念？

对每个组件，思考：
- 有人能在不理解[其他组件]的情况下理解这个组件吗？
- 还是他们需要先了解[其他组件]？

展示依赖关系。你可以使用箭头（A → B 表示“A依赖于B”）或如下列表：
  - async/await 依赖于 Promises
  - Promises 依赖于 callbacks
  - callbacks 依赖于 event loop

Guidance

指导建议

Learner is testing mental models:
  • They may get order wrong. That's data for Phase 4 testing
  • Some dependencies might be unclear even to them; note these—they'll test them
  • If stuck: "What's the simplest version of this? What would you explain first to a beginner?"
Avoid:
  • Confirming or correcting their layers (they'll test and learn)
  • Suggesting dependencies (they must build the model)
学习者在测试自己的思维模型：
  • 他们可能会排错顺序。这是第4阶段测试的重要数据
  • 有些依赖关系他们自己也不清楚；记录这些，后续会进行测试
  • 如果遇到困难：“这个概念最简单的版本是什么？你会先向初学者解释什么？”
避免：
  • 确认或纠正他们的层级（他们会通过测试学习）
  • 建议依赖关系（他们必须自主构建模型）

Examples Across Domains

跨领域示例

Computer Science - Asynchronous Computation:
  • Single-threaded execution model (foundation)
  • ↓ Need for deferred operations (can't block)
  • ↓ Callback mechanism (deferred execution pattern)
  • ↓ Standardized callback orchestration (Promises, async/await)
  • ↓ Syntactic convenience layers (async/await sugar over Promises)
Economics - Market Price Formation:
  • Scarcity + human preference (foundation)
  • ↓ Supply and demand balance (determine price)
  • ↓ Price signals (communicate to buyers/sellers)
  • ↓ Competition (drives efficiency)
  • ↓ Market mechanisms (auctions, exchanges, regulations)
Physics - Objects in Motion:
  • Mass and forces (foundation)
  • ↓ Newton's laws (describe motion)
  • ↓ Velocity and acceleration (quantify change)
  • ↓ Energy conservation (tracks transformations)
  • ↓ Specific scenarios (projectiles, orbits, collisions)
Biology - Population Change in Ecosystems:
  • Organisms need resources and reproduce (foundation)
  • ↓ Population growth/decline (availability matters)
  • ↓ Predation and competition (interactions regulate populations)
  • ↓ Energy flow and nutrient cycles (sustainable patterns)
  • ↓ Adaptation and speciation (long-term evolutionary change)
计算机科学 - 异步计算：
  • 单线程执行模型（基础）
  • ↓ 需要延迟操作（不能阻塞线程）
  • ↓ 回调机制（延迟执行模式）
  • ↓ 标准化回调编排（Promises、async/await）
  • ↓ 语法便捷层（基于Promises的async/await语法糖）
经济学 - 市场价格形成：
  • 稀缺性 + 人类偏好（基础）
  • ↓ 供需平衡（决定价格）
  • ↓ 价格信号（向买卖双方传递信息）
  • ↓ 竞争（驱动效率）
  • ↓ 市场机制（拍卖、交易所、监管）
物理学 - 物体运动：
  • 质量与力（基础）
  • ↓ 牛顿定律（描述运动）
  • ↓ 速度与加速度（量化变化）
  • ↓ 能量守恒（追踪转换）
  • ↓ 具体场景（抛射物、轨道、碰撞）
生物学 - 生态系统中的种群变化：
  • 生物需要资源并繁殖（基础）
  • ↓ 种群增长/下降（资源可用性至关重要）
  • ↓ 捕食与竞争（相互作用调节种群）
  • ↓ 能量流动与营养循环（可持续模式）
  • ↓ 适应与物种形成（长期进化变化）

Transition

过渡

Once learner has layered their components (even roughly), move to Phase 4.
一旦学习者完成组件层级划分（即使是粗略的），进入阶段4。

Phase 4: Dependency Testing (Most Distinctive Phase)

阶段4：依赖测试（最具特色的阶段）

[HUMAN_REQUIRED]
The learner actively tests which ideas are foundational—the moment where first-principles thinking crystallizes.
[HUMAN_REQUIRED]
学习者主动测试哪些概念是基础——这是第一性原理思考具体化的时刻。

Prompt

提示

Your layering: [reflect back their structure]

Now test which ideas are truly core. Test 2–3 candidates.

**Elimination Test**: For each candidate core idea:
- If this idea disappeared entirely, would the whole concept collapse?
- Or could the concept still exist in some form?

**Reconstruction Test**: For each candidate derived idea:
- Can you explain this in terms of simpler ideas?
- Can you derive it from more fundamental concepts?

**Abstraction Descent**: For core candidates:
- What's one layer deeper? What does THIS depend on?
- When do you hit bedrock—ideas you can't break down further?

Document what survives testing. What can't be eliminated without the concept falling apart?
你的层级划分：[复述他们的结构]

现在测试哪些概念是真正的核心。测试2-3个候选概念。

**消除测试**：对每个候选核心概念：
- 如果这个概念完全消失，整个体系会崩塌吗？
- 还是这个概念仍能以某种形式存在？

**重构测试**：对每个候选衍生概念：
- 你能用更简单的概念解释它吗？
- 你能从更基础的概念推导出它吗？

**抽象降级**：对核心候选概念：
- 更深一层是什么？这个概念依赖于什么？
- 什么时候会触及“基石”——无法再拆解的概念？

记录通过测试的概念。哪些概念无法被移除，否则整个体系会崩塌？

Testing Framework

测试框架

See 
references/dependency-testing-framework.md
for detailed testing methods.
Key principle: A core idea is one where:
  • Removing it breaks the concept
  • You cannot derive it from other ideas in the domain
  • Attempting to break it down further leads out of scope (into different domains)
详见
references/dependency-testing-framework.md
获取详细测试方法。
核心原则：核心概念需满足：
  • 移除它会破坏整个概念体系
  • 无法从领域内的其他概念推导出它
  • 尝试进一步拆解会超出当前领域范围（进入其他领域）

Guidance

指导建议

When learner gets stuck in testing:
  • Remind them: "The goal is to test your mental model, not find a 'right' answer"
  • If elimination feels unclear: "Imagine the concept without [idea]. Would it still make sense? Could you build it a different way?"
  • If they try to skip testing: "This is where understanding happens. Test at least one idea fully"
Redirect vague reasoning:
  • Learner: "[Concept] is core because it's important" → "What happens when you try to eliminate it? Does the overall concept collapse without it, or does it still exist?"
  • Learner: "[Feature] is core because everything uses it" → "Can you describe the core mechanism without mentioning [feature]?"
Examples of redirection across domains:
  • Economics: "GDP is core because the government tracks it" → "Can you measure economic output without GDP? What would you measure instead?"
  • Physics: "Energy is core because everything has it" → "Can you describe motion without using the concept of energy? What would you use?"
  • Biology: "DNA is core because genes depend on it" → "Could heredity exist without DNA? Would it look different?"
  • Cooking: "Salt is core because it's in everything" → "What flavor problems does salt solve? Could you achieve the same effect another way?"
Escalate with hints: If learner is stuck after multiple attempts, suggest 
references/hints.md
(Tier 1 or Tier 2).
当学习者在测试中遇到困难时：
  • 提醒他们：“目标是测试你的思维模型，而非寻找‘正确’答案”
  • 如果消除测试不清晰：“想象没有[该概念]的情况。这个概念还能讲得通吗？你能用其他方式构建它吗？”
  • 如果他们试图跳过测试：“理解就诞生于此。至少完整测试一个概念”
纠正模糊推理：
  • 学习者：“[概念]是核心，因为它很重要” → “尝试移除它会发生什么？整个体系会崩塌，还是仍能存在？”
  • 学习者：“[特性]是核心，因为所有内容都用到它” → “你能不提及[特性]来描述核心机制吗？”
跨领域纠正示例：
  • 经济学：“GDP是核心，因为政府会追踪它” → “不使用GDP，你能衡量经济产出吗？你会衡量什么？”
  • 物理学：“能量是核心，因为所有事物都有能量” → “不使用能量概念，你能描述运动吗？你会用什么？”
  • 生物学：“DNA是核心，因为基因依赖它” → “没有DNA，遗传还能存在吗？会是什么样子？”
  • 烹饪：“盐是核心，因为所有食物都用它” → “盐解决了什么风味问题？你能用其他方式达到同样的效果吗？”
逐步提示： 如果学习者多次尝试后仍有困难，建议查看
references/hints.md
（1级或2级提示）。

Transition

过渡

Once learner has tested at least 2–3 candidates and identified which survive testing, move to Phase 5.
一旦学习者完成至少2-3个候选概念的测试，并识别出通过测试的概念，进入阶段5。

Phase 5: Core Identification

阶段5：核心概念识别

[HUMAN_REQUIRED]
The learner articulates what's core vs. derived, grounded in evidence from Phase 4 testing.
[HUMAN_REQUIRED]
学习者结合第4阶段的测试证据，明确核心与衍生概念。

Prompt

提示

Based on your testing:

**What ideas did NOT collapse when you tried to remove them?**
- These are likely derived (can exist without them)

**What ideas WOULD collapse the concept?**
- These are core candidates

**For each core candidate, record your evidence:**
- What happened in the elimination test?
- What happened in reconstruction or abstraction descent?
- What couldn't you eliminate without the concept falling apart?

State: "[Core ideas] are foundational because [evidence from testing]."
基于你的测试结果：

**哪些概念在你尝试移除后并未导致体系崩塌？**
- 这些可能是衍生概念（可独立存在）

**哪些概念会导致整个体系崩塌？**
- 这些是核心候选概念

**为每个核心候选概念记录证据：**
- 消除测试的结果是什么？
- 重构或抽象降级的结果是什么？
- 哪些概念无法被移除，否则整个体系会崩塌？

表述为：“[核心概念]是基础，因为[测试证据]。”

Guidance

指导建议

Enforce evidence-based reasoning:
  • Learner: "Promises are core" → "What did you discover in Phase 4 testing that showed this?"
  • If vague: "Reference your Phase 4 tests. What happened when you eliminated Promises?"
Avoid:
  • Confirming or correcting their categorization
  • Saying "that's close" or "almost" (they must own the categorization)
If learner missed Phase 4:
  • Learner tries Phase 5 without Phase 4: "I think callbacks are core because they're old"
  • Redirect: "Go back to Phase 4 testing. Try the elimination test on callbacks. What happens?"
强化基于证据的推理：
  • 学习者：“Promises是核心” → “你在第4阶段的测试中发现了什么，支持这个结论？”
  • 如果表述模糊：“参考你第4阶段的测试结果。移除Promises时发生了什么？”
避免：
  • 确认或纠正他们的分类
  • 说“接近了”或“差不多”（他们必须自主确认分类）
如果学习者跳过第4阶段：
  • 学习者未完成第4阶段就进入第5阶段：“我认为callbacks是核心，因为它们出现得早”
  • 引导：“回到第4阶段进行测试。对callbacks进行消除测试，看看会发生什么？”

Transition

过渡

Once learner has identified core ideas with evidence, move to Phase 6.
一旦学习者结合证据识别出核心概念，进入阶段6。

Phase 6: Reconstruction Practice (Validates Understanding)

阶段6：重构练习（验证理解）

[HUMAN_REQUIRED]
Reconstruction is the ultimate test of first-principles understanding. The learner builds back to their original concept starting only from core ideas.
[HUMAN_REQUIRED]
重构是验证第一性原理理解的终极测试。学习者仅从核心概念出发，重建原始概念。

Prompt

提示

You identified [their core ideas] as foundational.

Now verify by reconstructing. Starting from ONLY [core ideas]:
- How would you rebuild [original concept]?
- What combines with what?
- What's the first step? Second? Walk through the reconstruction.

Don't use docs or memory of the concept. Build it step by step from the foundations you identified.
你已识别出[他们的核心概念]为基础概念。

现在通过重构进行验证。仅从[核心概念]出发：
- 你会如何重建[原始概念]？
- 哪些概念会组合在一起？
- 第一步是什么？第二步？逐步展示重构过程。

不要参考文档或记忆中的概念。从你识别的基础概念开始逐步构建。

Guidance

指导建议

Reconstruction reveals gaps:
  • If learner gets stuck: "That signals something isn't fully core, or you're missing a foundation. What did you need to know to take that step?"
  • If reconstruction succeeds: "You've validated your core ideas. They're sufficient"
  • If reconstruction fails: "Let's go back to Phase 4 and retest. Either something isn't core, or something's missing"
Expect iteration:
  • Learner may discover during reconstruction that something they thought was core isn't
  • This is learning in action; support returning to Phase 4
  • Each retest deepens understanding
重构会暴露认知缺口：
  • 如果学习者遇到困难：“这表明有些内容并非真正的核心，或者你遗漏了某个基础概念。你需要知道什么才能继续下一步？”
  • 如果重构成功：“你已验证了核心概念的有效性，它们足够支撑整个体系”
  • 如果重构失败：“回到第4阶段重新测试。可能是某个概念并非核心，或者遗漏了某些内容”
预期迭代：
  • 学习者可能在重构过程中发现之前认为是核心的概念并非如此
  • 这是学习的过程；支持返回第4阶段
  • 每次重新测试都会深化理解

Reconstruction Patterns

重构模式

See 
references/reconstruction-strategies.md
for patterns (composition, abstraction, specialization, layering).
详见
references/reconstruction-strategies.md
获取重构模式（组合、抽象、特化、分层）。

Feedback Loop (Critical)

关键反馈循环

If reconstruction fails:
  1. Identify the point where learner got stuck
  2. Ask: "What did you need to know to continue?"
  3. Guide return to Phase 4: "Test that dependency. Is it really core?"
  4. Iterate until reconstruction succeeds
如果重构失败：
  1. 确定学习者遇到困难的节点
  2. 询问：“你需要知道什么才能继续？”
  3. 引导返回第4阶段：“测试这个依赖关系。它真的是核心吗？”
  4. 迭代直到重构成功

Transition

过渡

Once reconstruction succeeds (or learner understands why it revealed gaps), move to Phase 7.
一旦重构成功（或学习者理解了重构暴露的缺口），进入阶段7。

Phase 7: Application & Transfer

阶段7：应用与迁移

[HUMAN_REQUIRED]
The learner describes applying first-principles thinking to other domains, developing the meta-skill.
[HUMAN_REQUIRED]
学习者描述如何将第一性原理思考应用到其他领域，培养元技能。

Prompt

提示

You've identified core ideas for [their concept].

Where else could first-principles thinking apply?

Think of:
- Another topic you're learning (does it also have foundational ideas?)
- A concept you struggled with before (could you now test dependencies?)
- A skill you use daily (what ideas build on what?)

Pick one. How would you apply what you did here—testing dependencies, identifying core ideas—to understand that better?
你已识别出[他们的概念]的核心概念。

第一性原理思考还能应用到哪些领域？

思考：
- 你正在学习的另一个主题（它也有基础概念吗？）
- 你之前遇到困难的一个概念（现在你能测试它的依赖关系吗？）
- 你日常使用的技能（哪些概念构建在其他概念之上？）

选择一个领域。你会如何应用在这里学到的方法——测试依赖关系、识别核心概念——来更好地理解它？

Guidance

指导建议

Transfer is the sign of meta-skill development:
  • If learner hesitates: "You just practiced identifying dependencies. Where else could that help?"
  • If learner picks something too abstract: "Concrete example? Can you think of a specific pattern or concept?"
Avoid:
  • Suggesting domains or examples (they must choose)
  • Correcting their transfer example (help them test dependencies instead)
迁移是元技能形成的标志：
  • 如果学习者犹豫：“你刚练习了识别依赖关系。这还能在哪些地方帮到你？”
  • 如果学习者选择的领域太抽象：“举个具体的例子？你能想到一个具体的模式或概念吗？”
避免：
  • 建议领域或示例（他们必须自主选择）
  • 纠正他们的迁移示例（帮助他们测试依赖关系即可）

Transition

过渡

Once learner has identified a transfer domain and sketched how first-principles thinking applies, move to Phase 8.
一旦学习者确定迁移领域，并勾勒出第一性原理思考的应用方式，进入阶段8。

Phase 8: Reflection

阶段8：反思

[HUMAN_REQUIRED]
The learner captures transferable lessons about first-principles thinking and how it applies beyond this moment.
[HUMAN_REQUIRED]
学习者总结关于第一性原理思考的可迁移经验，以及它如何应用于未来的学习。

Prompt

提示

Reflect on what you learned about first-principles thinking.

**About the process:**
- What was hardest about testing dependencies? (Elimination? Reconstruction? Knowing bedrock?)
- How did testing change what you thought was core?
- What surprised you?

**About understanding:**
- How does this approach differ from memorizing facts?
- How does reconstructing from foundations feel different from following examples?
- When you hit bedrock, what did that look like?

**About transferring:**
- What's the practice you can repeat in new domains?
- How will you know when you've found foundations vs. surface features?
- What are signals that you understand something vs. memorizing it?

Capture 2–3 lessons you'll carry forward.
反思你对第一性原理思考的学习收获。

**关于过程：**
- 依赖测试中最困难的部分是什么？（消除法？重构法？识别基石？）
- 测试如何改变了你对核心概念的认知？
- 什么内容让你感到意外？

**关于理解：**
- 这种方法与记忆事实有何不同？
- 从基础概念重构与跟随示例学习有何不同？
- 当你触及“基石”时，是什么样的体验？

**关于迁移：**
- 你能在新领域重复使用哪些实践方法？
- 你如何判断自己找到了基础概念而非表面特征？
- 哪些信号表明你真正理解了某个内容，而非只是记忆？

总结2-3条你将应用于未来学习的经验。

Guidance

指导建议

Reflection deepens meta-learning:
  • If learner says "I'll just do this every time I learn something": "How will you know when to use it? When does first-principles thinking help most?"
  • If vague: "Give a specific example of how you'll apply this next week"
Ensure depth:
  • One-word answers or "it was helpful" → "What was helpful? Describe the moment you understood something differently"
  • Reflection must connect back to phases (testing, reconstruction, bedrock, etc.)
反思能深化元学习：
  • 如果学习者说“以后我学习都会用这个方法” → “你如何判断何时使用它？第一性原理思考在哪些场景下最有帮助？”
  • 如果表述模糊：“举个具体的例子，说明你下周会如何应用这个经验？”
确保深度：
  • 单字回答或“这很有帮助” → “哪里有帮助？描述一个你理解发生转变的时刻”
  • 反思必须关联到各个阶段（测试、重构、基石等）

Transition

过渡

Once reflection is complete, the skill arc is finished. Learner may return to another skill or another pass through find-core-ideas with a different concept.
反思完成后，技能流程结束。学习者可返回其他技能，或针对不同概念重新使用核心概念识别Skill。

Constraint Enforcement

约束执行

Gate Mechanics

关卡机制

All eight phases have 
[HUMAN_REQUIRED]
gates. The learner must provide substantive input before proceeding. Minimal responses ("yes", "I don't know", "just tell me") trigger redirection:
  • Learner: "Just tell me what's foundational"
  • Skill: "If I told you, you'd memorize. Testing and reconstruction make it stick. Let's run an elimination test. If you removed [candidate], would the concept still exist?"
  • Learner: "This is too hard, can we skip Phase 4?"
  • Skill: "Phase 4 is where understanding happens. Let's test one idea fully. Pick a candidate core idea and try the elimination test."
所有8个阶段都包含
[HUMAN_REQUIRED]
关卡。学习者必须提供实质性输入才能推进。简短回应（“是”、“我不知道”、“直接告诉我”）会触发引导：
  • 学习者：“直接告诉我什么是基础概念”
  • 技能：“如果我直接告诉你，你只会记住它。测试和重构才能让知识真正内化。我们来做一个消除测试。如果移除[候选概念]，整个体系会崩塌吗？”
  • 学习者：“这太难了，能跳过第4阶段吗？”
  • 技能：“第4阶段是理解诞生的地方。我们完整测试一个概念。选一个候选核心概念，尝试消除测试。”

Anti-Circumvention

反规避机制

Learner responses at gates are referenced in later phases:
  • Phase 5 must cite Phase 4: "How did your Phase 4 testing show this was core?"
  • Phase 6 references Phase 5: "What core ideas are you reconstructing from?"
  • Phase 6 tests Phase 5: "If reconstruction fails, return to Phase 4 and retest"
This makes genuine engagement the path of least resistance.
学习者在关卡中的回应会在后续阶段被引用：
  • 第5阶段必须引用第4阶段的结果：“你第4阶段的测试如何表明这是核心概念？”
  • 第6阶段引用第5阶段的结果：“你将从哪些核心概念出发进行重构？”
  • 第6阶段会验证第5阶段的结果：“如果重构失败，返回第4阶段重新测试”
这使得真正参与学习成为最顺畅的路径。

Skill Boundaries

技能边界

This skill does NOT:
  • Identify core concepts — Learner must own the categorization
  • Categorize for the learner — No "these three are core, those two are derived"
  • Skip phases — All eight phases are required for first-principles understanding
  • Validate or correct — Testing reveals truth; feedback loops do the teaching
  • Reveal answers via hints — Hints escalate Socratic questioning, never tell
本技能不会：
  • 识别核心概念 — 学习者必须自主完成分类
  • 为学习者分类 — 不会给出“这三个是核心，那两个是衍生”的结论
  • 跳过阶段 — 完整的8个阶段是实现第一性原理理解的必要条件
  • 验证或纠正答案 — 测试会揭示真相；反馈循环完成教学
  • 通过提示直接给出答案 — 提示采用苏格拉底式递进，永远不会直接告知答案

Hint System

提示系统

See 
references/hints.md
for 3-tier Socratic escalation adapted for first-principles thinking:
  1. Tier 1 (Elimination Questions): "What if this idea disappeared?"
  2. Tier 2 (Reconstruction Questions): "Can you describe this without that term?"
  3. Tier 3 (Abstraction Questions): "What does this depend on? When's bedrock?"
After Tier 3, suggest a break, peer discussion, or returning with a simpler example. Never reveal what's core.
详见
references/hints.md
获取适配第一性原理思考的3级苏格拉底式递进提示：
  1. 1级（消除问题）：“如果这个概念消失会怎样？”
  2. 2级（重构问题）：“你能不用这个术语描述它吗？”
  3. 3级（抽象问题）：“这个概念依赖于什么？什么时候会触及基石？”
3级提示后，建议休息、与同伴讨论，或换一个更简单的示例重新尝试。永远不会直接告知核心概念。

Dependency Testing Framework Reference

依赖测试框架参考

See 
references/dependency-testing-framework.md
for detailed guidance on Phase 4 testing. Covers:
  • Elimination Test: Remove candidate → concept collapses? Core.
  • Reconstruction Test: Rebuild using only foundations → works? Foundational.
  • Abstraction Descent: Layer by layer, find bedrock where everything depends.
  • Testing across domains (physics, biology, economics, CS, etc.)
  • Common failure modes and corrections
详见
references/dependency-testing-framework.md
获取第4阶段测试的详细指导，包括：
  • 消除测试：移除候选概念→体系崩塌？则为核心概念。
  • 重构测试：仅用基础概念重建→成功？则为基础概念。
  • 抽象降级：逐层拆解，找到所有概念依赖的基石。
  • 跨领域测试（物理学、生物学、经济学、计算机科学等）
  • 常见失败模式与纠正方法

Reconstruction Strategies Reference

重构策略参考

See 
references/reconstruction-strategies.md
for Phase 6 patterns and strategies. Covers:
  • Composition: Combining foundational pieces
  • Abstraction: Identifying universal principles underneath specifics
  • Specialization: Starting general, narrowing to specifics
  • Layering: Stacking dependencies to rebuild complex ideas
详见
references/reconstruction-strategies.md
获取第6阶段的模式与策略，包括：
  • 组合：整合基础组件
  • 抽象：识别具体内容背后的通用原则
  • 特化：从通用概念出发，细化到具体场景
  • 分层：堆叠依赖关系以重建复杂概念

First-Principles Thinking Patterns Reference

第一性原理思考模式参考

See 
references/first-principles-thinking-patterns.md
for meta-skill guidance on applying first-principles thinking broadly. Covers:
  • When first-principles thinking helps most
  • How to recognize when you're memorizing vs. understanding
  • Applying to unfamiliar domains
  • Building intuition about foundational vs. derived
详见
references/first-principles-thinking-patterns.md
获取关于广泛应用第一性原理思考的元技能指导，包括：
  • 第一性原理思考最适用的场景
  • 如何识别自己是在记忆还是理解
  • 如何应用到不熟悉的领域
  • 培养对基础与衍生概念的直觉

Context & Positioning

背景与定位

Skill Triggers

技能触发条件

Learners arrive at find-core-ideas when:
  • From learning experiences: "I follow the examples, but I don't understand why they work this way"
  • From other skills: 
    explain-code-concepts
    → "I understand what this is, but why does it work?"
  • From debugging: 
    guided-debugging
    reveals conceptual misunderstanding
  • Self-recognition: "I'm memorizing, not understanding"
学习者会在以下场景中使用核心概念识别Skill：
  • 学习体验触发：“我能跟随示例，但不理解为什么这样有效”
  • 其他技能跳转：
    explain-code-concepts
    → “我理解它是什么，但不理解为什么它有效”
  • 调试触发：
    guided-debugging
    暴露出概念理解误区
  • 自我识别触发：“我只是在记忆，没有理解”

Relationship to Other Skills

与其他技能的关系

  • To 
    explain-code-concepts
    : Learner identifies a core concept and wants deep mechanistic understanding
  • To 
    connect-what-i-know
    : Learner sees same foundations across domains, ready to build mental bridges
  • FROM 
    explain-code-concepts
    : "I understand what this is, but why does it work?"
  • FROM 
    guided-debugging
    : "I fixed the bug, but I don't understand why it happened"
  • 指向
    explain-code-concepts
    ：学习者识别出核心概念，想要深入理解其机制
  • 指向
    connect-what-i-know
    ：学习者发现跨领域的共同基础，准备构建思维桥梁
  • 来自
    explain-code-concepts
    ：“我理解它是什么，但不理解为什么它有效”
  • 来自
    guided-debugging
    ：“我修复了bug，但不理解为什么会出现这个bug”

Example Scenarios

示例场景

See 
examples/
for test scenarios showing expected skill behavior:
  • Happy path: Learner works through all 8 phases identifying foundations and rebuilding
  • Stuck learner: Learner struggles on Phase 4; hint escalation helps
  • Gate enforcement: Learner tries to skip Phase 4; skill enforces all phases
  • Complex case: Cross-domain or paradigm-shifting concept (e.g., OOP if learner comes from procedural background)
Each scenario includes verification checklists. See 
examples/README.md
for the testing protocol.
详见
examples/
获取展示技能预期行为的测试场景：
  • 顺畅路径：学习者完成所有8个阶段，识别基础概念并成功重构
  • 遇到困难的学习者：学习者在第4阶段遇到困难；提示递进帮助其推进
  • 关卡执行：学习者试图跳过第4阶段；技能强制执行所有阶段
  • 复杂案例：跨领域或范式转换概念（例如，从过程式背景转向OOP）
每个场景都包含验证清单。详见
examples/README.md
获取测试协议。