---

name: book-docusaurus description: Scaffold, structure, and deploy the Physical AI textbook in Docusaurus with book-aware content and RAG-ready exports. Use when creating or updating the Docusaurus site, adding chapters, configuring sidebar, or deploying to GitHub Pages/Vercel.

Book Docusaurus Skill

Instructions

1. Scaffold the site

   • Ensure Node >=18 is installed
   • Run npx create-docusaurus@latest physical-ai-book classic in project root or /docs
   • Configure docusaurus.config.js with site metadata, GitHub Pages URLs, i18n (en default)

2. Structure content

   • Build sidebars.js for Quarter overview, Modules 1-4, Capstone, Assessments, Hardware kits, Cloud option
   • Create MDX stubs per module/week with learning outcomes and tasks
   • Add capstone outline

3. Authoring affordances

   • Add MDX components for callouts, checklists, hardware tables, code blocks
   • Enable search (Algolia DocSearch placeholder) and local search plugin for dev

4. Deploy

   • Add GitHub Actions workflow for GH Pages (npm ci, npm run build, npm run deploy)
   • Document Vercel deploy steps (import repo, build command npm run build, output build)

5. RAG-readiness

   • Enforce frontmatter fields: title, description, module, week, tags
   • Keep headings semantic (h2 for weeks, h3 for sections)
   • Avoid heavy client-side rendering for core text
   • Export ingestion guidance: markdown path glob, ignore build/static

Examples

# Create new chapter
mkdir -p docs/module-1/week-1
cat > docs/module-1/week-1/intro.mdx << 'EOF'
---
title: Introduction to Physical AI
description: Overview of embodied intelligence and humanoid robotics
module: 1
week: 1
tags: [physical-ai, robotics, introduction]
---

# Introduction to Physical AI

Content here...
EOF

# Build and test locally
npm run build
npm run serve

Definition of Done

• npm run build passes; site renders outline and sample content
• Sidebar matches course hierarchy; links valid
• GH Pages workflow present; deploy instructions written
• Content annotated with frontmatter and semantic headings suitable for chunking