Concept Taxonomy
The 570 concepts in the learning graph are organized into
14 categories. Each category has a short TaxonomyID used in the
TaxonomyID column of learning-graph.csv and as the
group key in the graph viewer legend.
Categories were chosen so that no single category dominates the graph. The largest category holds under 12% of all concepts, well below the 30% ceiling used as a quality threshold for this project.
Categories
Foundations of AI and Intelligent Books
TaxonomyID: FOUND
Core ideas a reader needs before anything else: what a large language model is, what tokens and context windows are, how an AI coding agent differs from a chat interface, and what makes a textbook "intelligent" across the five levels of textbook intelligence.
Agent Skill Architecture
TaxonomyID: SKARCH
The structure of a skill itself — the SKILL.md file, the frontmatter contract,
the directory layout, progressive disclosure and its three loading budgets, and
the meta-skill routing pattern that keeps a large library under the loading limit.
Skill Development and Portability
TaxonomyID: SKDEV
Building, testing, installing, and distributing skills, plus everything involved in making one skill library run across Claude Code, Codex, Gemini, Cursor, and Copilot without forking it.
Token Optimization and Measurement
TaxonomyID: TOKEN
Treating tokens as an engineering budget: plan limits and usage windows, the cost of serial versus parallel sub-agents, file layout as a token strategy, and the hooks and logs used to measure what each skill actually consumes.
Python Tooling and Automation
TaxonomyID: PYTOOL
The scripts that do the deterministic work — parsing, validating, scaffolding, counting, and converting — together with the general Python practices that make those scripts safe to run repeatedly.
Course Design and Pedagogy
TaxonomyID: CDESIGN
Course descriptions, Bloom's Taxonomy and its six cognitive levels, learning outcomes, reading level, instructional scaffolding, and the pedagogical mascot conventions used to guide readers.
Learning Graphs
TaxonomyID: LGRAPH
Concepts, learning dependencies, and the directed acyclic graph they form: enumeration, dependency mapping, quality metrics, taxonomy assignment, and the JSON format that drives the interactive graph viewer.
Chapter and Content Generation
TaxonomyID: CONTENT
Turning a learning graph into chapters — structure design, concept assignment, section organization, admonitions, math support, and the specification blocks that later drive diagram and simulation generation.
Supporting Content
TaxonomyID: SUPPORT
The material that surrounds the chapters: ISO 11179-compliant glossaries, FAQs and their chatbot exports, quizzes and distractor quality, and curated reference lists that credit the authors behind influential explanations.
Interactive Simulations
TaxonomyID: MSIM
MicroSims end to end — directory structure and file separation, metadata schemas, the visualization libraries the generator routes between, iframe embedding and height management, and the quality and screenshot utilities that maintain them.
Images, Infographics, and Media
TaxonomyID: MEDIA
Everything generated as pixels or audio: text-to-image models and their fabrication risks, the verified infographic pipeline, interactive overlay engines, freely-licensed image sourcing, slide decks, and text-to-speech.
Domain-Specific Skill Extension
TaxonomyID: DOMAIN
Extending a subject-neutral library into a specific field, using the beginning-electronics case study: circuit schematic generation, breadboard simulations, and rubric-driven hands-on lab evaluation.
Platform, Tooling, and Deployment
TaxonomyID: PLATFORM
MkDocs and the Material theme, site features, the editor and terminal, Git and GitHub, deployment to GitHub Pages, and analytics registration.
Publishing, Metrics, and Promotion
TaxonomyID: PUBLISH
Measuring a finished book and announcing it: the canonical book metrics hub, README generation, LinkedIn posts and carousels, press releases, and the session logs that record how the book was built.