Understanding

Understanding is a public Codex skill repository for lijie, a learning-oriented skill that helps an AI assistant turn unfamiliar material into clear, testable knowledge structure.

Understanding 是 lijie 这个 Codex skill 的公开仓库。它的目标不是简单压缩文本，而是帮助 AI 助手把陌生知识拆成「讲得清、拆得开、用得上、能验证」的结构。

What This Skill Does / 这个 Skill 的作用

lijie guides Codex to process learning material through three lenses:

• Feynman explanation: explain simply first, then precisely.
• First principles: separate primitives, assumptions, mechanisms, constraints, examples, and consequences.
• Knowledge structure mapping: label sequences, parallel sets, hierarchies, dependencies, networks, many-to-one relations, one-to-many relations, closed loops, and open loops.

lijie 会引导 Codex 从三个角度处理学习材料：

• 费曼解释：先用简单语言讲清楚，再给出精确版本。
• 第一性原理：区分底层概念、事实、假设、机制、限制、例子和结果。
• 知识结构映射：标出顺序、并列、层级、依赖、网络、多对一、一对多、闭环和开环。

When To Use It / 什么时候用

Use it when you want to understand, teach, summarize, or internalize:

• a concept, paper, article, book, course, video, or documentation page
• a technical system, workflow, business model, mental model, or research domain
• dense notes that need structure rather than compression
• hidden relationships between ideas, especially dependencies, feedback loops, tradeoffs, and prerequisites

当你想学习、解释、总结或内化这些内容时，可以使用它：

• 概念、论文、文章、书籍、课程、视频或文档
• 技术系统、工作流、商业模型、思维模型或研究领域
• 信息很密但结构不清的笔记
• 概念之间的隐藏关系，尤其是依赖、反馈闭环、权衡和前置知识

Quick Start / 快速开始

Clone this repository:

git clone https://github.com/aDragon0707/understanding.git
cd understanding

Install the skill into your local Codex skills directory:

Copy-Item -Recurse -Force .\skill\lijie C:\Users\<you>\.codex\skills\

Then invoke it in Codex:

Use $lijie to explain reinforcement learning from first principles, then map its knowledge structure.

克隆仓库：

git clone https://github.com/aDragon0707/understanding.git
cd understanding

安装到本地 Codex skills 目录：

Copy-Item -Recurse -Force .\skill\lijie C:\Users\<you>\.codex\skills\

然后在 Codex 中调用：

用 $lijie 解释这篇论文：先给我费曼解释，再拆第一性原理，最后列出掌握检查题。

Example Prompts / 示例提示词

Use $lijie to summarize this article, then map the knowledge structure and open loops.

Use $lijie to turn these notes into a study plan with mastery checks.

用 $lijie 把这个商业模式拆成：底层假设、关键机制、依赖关系、反馈闭环和失败模式。

用 $lijie 学习这个新概念：先讲人话，再讲专业版，最后给我自测题。

Repository Structure / 仓库结构

.
|-- README.md
|-- CHANGELOG.md
|-- LICENSE
|-- docs/
|   |-- INTRO.en.md
|   `-- INTRO.zh-CN.md
`-- skill/
    `-- lijie/
        |-- SKILL.md
        |-- agents/
        |   `-- openai.yaml
        `-- references/
            `-- structure-framework.md

Skill Package / Skill 包内容

The installable skill lives in skill/lijie/.

• skill/lijie/SKILL.md: the main skill instructions and trigger metadata
• skill/lijie/agents/openai.yaml: UI-facing metadata for Codex
• skill/lijie/references/structure-framework.md: a reference for mapping knowledge relationships and feedback loops

真正可安装的 skill 位于 skill/lijie/。

• skill/lijie/SKILL.md：主说明和触发元数据
• skill/lijie/agents/openai.yaml：Codex 界面使用的展示元数据
• skill/lijie/references/structure-framework.md：知识关系和反馈闭环的结构映射参考

More Docs / 更多文档

• English introduction
• 中文介绍
• Changelog

Design Philosophy / 设计理念

Good learning is not just shorter text. It is structure, mechanism, feedback, and recall.

好的学习不是把文字变短，而是把知识变成结构、机制、反馈和可回忆的能力。

lijie therefore pushes the assistant to answer:

• What is the core question?
• What are the irreducible pieces?
• How do the pieces generate the result?
• What depends on what?
• Where is the feedback loop?
• How can the learner prove they can use it?

所以 lijie 会推动助手回答：

• 这个知识到底在回答什么核心问题？
• 哪些是不可再拆的底层零件？
• 这些零件如何生成最终结论？
• 哪些概念依赖哪些概念？
• 哪里有反馈闭环，哪里只是开环？
• 怎样证明学习者真的会用？

License / 许可证

MIT License. See LICENSE.