build-course

name	build-course
description	Explore a target codebase and generate a complete course for it in the M:\Courses platform. Use when the user wants to create a course about another project, codebase, or tool — e.g. "build a course for M:\ai-dev-net", "create a course about this repo", "make a course on X".

Build Course

Explore a target codebase, design a curriculum, then write all course files into M:\Courses\courses\.

Before writing a single lesson, read the authoring reference: M:\Courses\courses\How to Create Courses\ — every lesson in that course is a spec. Follow it exactly for structure, callout syntax, quiz frontmatter, and quality rules.

Inputs

The user provides:

• Target codebase path — the repo to learn from (e.g. M:\ai-dev-net)
• Audience (optional) — who will take the course; default to "developers"
• Depth (optional) — "intro" (3–4 modules) or "deep" (6–8 modules); default "intro"

Phase 1 — Read the Authoring Guide

Spawn an Explore subagent against M:\Courses\courses\How to Create Courses\ and extract:

• Required file/folder structure and naming conventions
• Lesson writing pattern (4-part structure, Key Takeaway rule)
• Callout syntax and available types
• Quiz frontmatter schema
• Content quality checklist

Hold this as your authoring rulebook for the rest of the skill.

Phase 2 — Explore the Target Codebase

Spawn a second Explore subagent against the target path. Ask it to return:

1. What the project is — purpose, problem it solves, primary tech stack
2. Key concepts & mental models — domain terms, core abstractions, architecture patterns
3. User-facing features — what someone does with it day to day
4. Getting-started path — setup steps, entry points, first meaningful thing a learner can do
5. Non-obvious gotchas — anything that trips up newcomers

Search across README files, docs/, src/ entry points, config files, and example/demo directories.

Run Phases 1 and 2 in parallel.

Phase 3 — Design the Curriculum

Using both outputs, design the course before writing any files:

1. Course metadata

   • courseId: slugified title (lowercase, hyphens, no spaces)
   • title, subtitle, audience

2. Module plan (3–4 for intro, 6–8 for deep)

   • Each module: a thematic cluster of 4–6 lessons + 1 quiz
   • Module overview title pattern: # Module N Overview: {Theme}
   • Progression: orient → understand → apply → assess

3. Lesson plan per module

   • Files numbered from 1-module-overview.md through N-{topic}.md
   • Final file per module: the quiz (e.g. 7-{theme}-quiz.md)
   • File names: lowercase, hyphen-separated, numeric prefix

4. Quiz design (5–8 questions per module)

   • Four options each; one correct answer that exactly matches an option string
   • Include a brief explanation per question

Output the plan as a brief list and confirm the structure is sound before writing.

Phase 4 — Write the Course Files

Courses root: M:\Courses\courses\

Folder layout

M:\Courses\courses\{Course Folder Name}\
├── course.json
├── 1-intro.md
└── modules\
    ├── 01\
    │   ├── 1-module-overview.md
    │   ├── 2-{topic}.md
    │   ├── ...
    │   └── N-{theme}-quiz.md
    ├── 02\
    └── ...

• Course folder name = title as written (spaces preserved), e.g. AI Dev Studio Fundamentals
• Module folders: zero-padded two digits (01, 02, … 10)

course.json

{
  "courseId": "{{courseId}}",
  "title": "{{title}}",
  "subtitle": "{{subtitle}}",
  "audience": "{{audience}}"
}

1-intro.md frontmatter

---

courseId: {{courseId}}
title: {{title}}
subtitle: {{subtitle}}
audience: {{audience}}
---

Lesson template (follow the 4-part structure from the authoring guide)

Every content lesson must end with a ## Key Takeaway section — one crisp sentence.

Quiz frontmatter

---

lessonType: quiz
passThreshold: 80
maxAttempts: 2
resetScopeOnFail: module
questions:
  - prompt: "{{Question?}}"
    options:
      - "{{A}}"
      - "{{B}}"
      - "{{C}}"
      - "{{D}}"
    answer: "{{A}}"
    explanation: "{{Why A is correct}}"
---

Quality Checklist

• Read How to Create Courses authoring guide before writing any content
• courseId in course.json matches frontmatter in 1-intro.md
• Every module has a 1-module-overview.md
• Every module ends with a quiz file
• Every quiz answer exactly matches one of its options strings
• Every lesson ends with ## Key Takeaway
• Callouts use > [!TYPE] syntax (NOTE, TIP, WARNING, SUCCESS, INFORMATION, IMPORTANT)
• Content references real artefacts from the target codebase (file names, functions, concepts)
• No module has more than 8 lesson files
• Run npm run build from M:\Courses\site to confirm no loading errors