Teach Yourself Skill
Skill for building study courses: from task and source definition through chapter checks to publishing in Markdown and HTML.
- 125 ★ on the repository
- Last commit
Minimal access
Why this level
- Writes course files into the working folder and runs a Node.js check script.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/teach-yourself-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Writes course files into the working folder and runs a Node.js check script.
This is third-party code. Review the repository files before installing.
What it does
Teach Yourself Skill walks an agent through making a course as a product with a verifiable result. The agent first records the course task, audience and sources in workspace files, then drafts a chapter plan, writes the chapters and builds them into Markdown and HTML. The skill ships six chapter-body types, section templates, course scoring criteria and the script `scripts/validate-course.mjs`. The dependency-free script checks folder structure, closed code blocks, HTML and local links. A long course runs in batches of three to four chapters: after each batch the text goes through a language check, the script and scoring against the criteria. The repository includes a finished nine-chapter example about high-quality AI code review.
Who it is for. Course authors and instructional designers who build learning material with an agent and want chapters checked against a list.
Good fit when
- A beginner course is needed for an internal tool, with code and docs around it.
- A lesson series keeps growing and chapters should be checked against shared criteria.
- An existing course must be rewritten or extended without losing its structure.
- Before publishing you want a script run over links and HTML.
Not a fit when
- You need one short cheat sheet, not a course with a plan and checks.
- You have no sources: the skill grounds claims in material you supply.
- The course language is not Chinese: instructions and the language check are written for Chinese text.
Example request
Use teach-yourself-skill to create a beginner course on our API: first the plan and source list, then the first chapter.Limitations
The full cycle needs a second skill, humanizer-zh, placed in the same skills folder as this one. Without it the plan and script checks work, but the language check cannot be skipped if you want the whole loop. Instructions and criteria are written in Chinese and tuned for Chinese text. The check script runs on Node.js. The repository has no license, so confirm reuse terms with the author. A course of dozens of chapters takes many model requests.
How to disable. Delete the `teach-yourself-skill` folder from the agent's skills directory. Courses you created live elsewhere and stay in place.
FAQ
How do I install Teach Yourself Skill for Codex?
Run `npx skills@latest add chokcoco/teach-yourself-skill --agent codex --global --yes`. In a task, call the skill by name as `$teach-yourself-skill`.
Why is humanizer-zh needed?
It checks chapter text for bureaucratic and machine-sounding phrasing. The skill treats it as a required dependency, and there is no general way to install another skill automatically, so you place the folders side by side by hand.
How do I check a course without a model?
Run `node scripts/validate-course.mjs examples/high-quality-ai-code-review` or the path to your own course. The script needs no dependencies.
What stages does the work have?
Five stages: task and sources, plan, two benchmark chapters, remaining chapters in batches, and a check of every chapter.
What is inside the skill folder?
`SKILL.md` with the workflow, `scripts/` with the checks, `references/` with criteria, `templates/` with chapter templates and `examples/` with a finished course.
Discussion
No questions or reviews yet.
Related
SkillsA Claude Code skill that builds an isometric explainer: a vehicle travels the map and each stop runs one real simulation step.
A skill for interface copy: errors, buttons, empty states and forms against four standards, scored with a checklist.
A set of 25 skills for Codex and Claude Code: articles, research, Remotion video, interfaces, GitHub and pre-release verification
Marketing Skills for AI agents
Marketing Skills for AI Agents
About fifty skills for CRO, copywriting, SEO, analytics, ads and launches, built for coding agents
Marketing & SEO Skills
Marketing & SEO Skills for AI Agents
A library of hundreds of marketing and SEO skills for agents: content, page types, channels, ads and strategies
Twenty-two skills for marketing and engineering: copy, social posts, Chrome extensions, PromQL, research and skill checks
README in short
The repository is laid out as a skill folder: `SKILL.md`, check scripts, references, chapter templates and an example course. Work is split into stages with checkpoints, and text edits invalidate earlier check reports.
README badge
Original SKILL.md
--- name: teach-yourself-skill description: 面向不同年龄、基础、职业和学习目标，规划、生成、质检并迭代通用型课程或系列教程。用于从任务定义与可信资料出发，按环境准备、语言基础、框架应用、原理诊断、综合实践、开篇总结等正文类型动态选择写法与讲师风格，设计可长可短的章节体系，生成 Markdown、HTML、练习、项目和参考资料，并通过确定性脚本、课程评价量规、humanizer-zh、回归检查与人工审核形成质量闭环；也用于重构、扩写、审阅或发布已有课程。 --- # Teach Yourself Skill 把课程当作一个有来源、有验收标准、可持续迭代的教学产品。不要把它压缩成一组固定长度的短课，也不要把“生成完文件”视为完成。 ## 先读取约束 开始前完成以下动作： 1. 查找并遵守仓库根目录及目标课程目录中的 `AGENTS.md`。 2. 读取目标课程 workspace 中已有的 `MISSION.md`、`RESOURCES.md`、`COURSE-GENERATION-SPEC.md`、`COURSE-BLUEPRINT.md` 和 `NOTES.md`。 3. 读取目标目录已有的生成脚本、课程目录、样章、参考资料和质量报告。不要覆盖用户改动。 4. 生成或改写中文学生可见文本前，完整读取已安装的 [Humanizer-zh](https://github.com/op7418/Humanizer-zh) 的 `humanizer-zh/SKILL.md`，并执行本文的 Humanizer 闸门。安装约定见 [humanizer 依赖](references/humanizer-dependency.md)。 5. 进入第二阶段语义质检前，完整读取 references/course-rubric.md。不要只凭本文件中的摘要自行发明评分标准。 如果课程 workspace 尚不存在，先根据用户目标创建最小任务定义。信息不足但不影响方向时，记录假设并继续；受众、课程目标、交付格式或事实来源的缺失会显著改变产物时，先向用户确认。 ## 建立课程任务定义 先把模糊的“做一门课”改写成可验收的任务。至少记录： - 课程为什么存在，以及学习完成后要解决什么问题 - 核心受众与必要的次级受众 - 学习者的前置知识、语言、年龄或职业背景 - 预期学习成果，以及学习者如何证明自己会了 - 可投入时间、学习节奏和使用场景 - 内容范围、明确不讲的内容和风险边界 - 每个章节的有效字符数字数要求，默认值设定为 12000，允许用户在课程初始化时候修改该值 - 交付格式、运行环境、发布方式与无障碍要求 - 来源政策、时效要求和允许使用的工具 - 人工审核者、发布权限和停止条件 把课程任务定义写入或更新 `MISSION.md` 与 `COURSE-GENERATION-SPEC.md`。已有任务定义优先于通用规则；发现冲突时，不静默选择其中一份。 ## 采用可变粒度，而不是固定短课 根据学习目标和认知负荷决定课程与章节大小，不设置统一篇数、字数或时长。 - 只有一个窄目标、依赖很少时，生成精简教程或单章课程。 - 目标包含多个阶段性能力时，生成模块化课程，并显式标出先修关系。 - 主题需要完整理论、案例、实践和复盘时，允许章节较长；用清晰分段和导航控制阅读负担。 - 同一章出现多个可独立验收的学习成果，或学习者必须跨越明显的前置知识台阶时，拆章。 - 两章分别无法形成完整解释或有效练习时，合并。 - 不为追求“丰富”强行加入代码、图表、案例、术语框或练习；只保留能帮助达成学习成果的材料。 - 不以字数代替深度。深度由机制解释、边界、证据、迁移练习和常见错误共同体现。