Skip to content

Chapter 2 — Change Log

A topic-organized record of edits to this chapter and the supporting files it depends on. Maintained by hand from session transcripts. Use this when you need to understand why the chapter looks the way it does, not just what is in it now.


Session 2026-04-11 — Initial generation, image plan, story split, voice cleanup, emoji audit

This was the first content generation pass for Chapter 2. The chapter started the day as a stub containing only the title, summary, concept list, prerequisites, and a TODO: Generate Chapter Content line. By the end of the session it was a published-ready chapter with a companion mini graphic novel and several supporting infrastructure changes captured below.

1. Initial chapter generation (economy mode)

What changed: Replaced the TODO: Generate Chapter Content line with the full chapter body.

Method: Ran the chapter-content-generator skill in economy mode — a strict-constraint prompt that forbids plan mode, subagents, extended thinking, and unnecessary file reads. The skill was instructed to read exactly four files in order:

  1. The chapter stub (this file's earlier state)
  2. Chapter 1 (docs/chapters/01-welcome-to-digital-world/index.md) as the voice/structure template
  3. docs/learning-graph/mascot-test.md for verbatim mascot admonition syntax
  4. CLAUDE.md for the Grade 5 style guide

What it produced: A ~2,300-word chapter covering all 15 concepts from the "Concepts Covered" list in dependency order, with:

  • YAML frontmatter (title, description, generated_by, date, version)
  • An opening scenario starring Jordan, a fifth-grade boy, on the playground at recess
  • Five major sections that defined every concept using the bold-and-define pattern
  • Five Maka admonitions (welcome, thinking, tip, warning, celebration), no back-to-back, with the trusted-adult rule kept in plain prose
  • One MicroSim spec block (Pause, Think, Act decision tool) inside a <details markdown="1"> block
  • One comparison table (opportunities / responsibilities / rights)
  • A numbered list breaking down pause-think-act
  • A 15-item recap list

Why Jordan and not Maya: Chapter 1 already used Maya as its named fictional student. The CLAUDE.md style guide calls for varying names across chapters, so Jordan was chosen for Chapter 2. Maintaining a roster of recurring fictional kids will help give the textbook a sense of place over time.

2. Image and diagram specs added

What changed: Added five #### Diagram: blocks to the chapter, each containing a detailed AI-image-generator prompt inside a <details markdown="1"> block. The five images were:

  1. Jordan on the Playground — opening scenario illustration
  2. The Three-Legged Stool — concept metaphor for opportunities, responsibilities, rights
  3. Standing at the Digital Threshold — concept metaphor for the doorway / "moment before you tap"
  4. The Four-Wheeled Wagon — concept metaphor for the four habits (etiquette, ethics, law, safety)
  5. Jordan Walks Away — closing scenario resolution

Why: Visual reinforcement of metaphors and the narrative bookends. Each spec block captured a full text-to-image prompt with composition, characters, palette, and style notes so the prompts could be regenerated cleanly later.

3. Split: concept images stay inline, Jordan narrative becomes a graphic novel

What changed: The two narrative Jordan images (#1 and #5 above) were removed from the chapter and reborn as a separate 8-panel mini graphic novel at docs/stories/jordan-one-second-choice/index.md. The three concept-metaphor images (#2, #3, #4) remained inline.

Why: A two-image teaser cannot carry a full narrative arc. Jordan's story has at least eight beats (invitation → video → ask → knot in stomach → pause → think → act → walk to a trusted adult), and bundling them into a standalone graphic novel:

  • Lets the visual story breathe and teach the pause-think-act habit through showing, not just naming.
  • Makes the story a reusable artifact a teacher can project or a student can revisit outside the chapter.
  • Aligns with the textbook's stated direction in CLAUDE.md ("recurring fictional characters can appear across chapters; the textbook plans for graphic-novel stories").
  • Tightens the chapter prose by moving the dramatic narrative weight to the story.

The concept-metaphor images stayed inline because they only make sense next to the paragraph that introduces the metaphor. A student reading "like the three legs of a stool" benefits from seeing the stool in the same eyeful; linking to a separate page would break the metaphor loop.

Mini graphic novel structure: Generated via the story-generator skill, adapted for a fictional contemporary character (no historical-figure references, no Wikipedia citations). Final shape: cover image + 8 panels (Recess Begins → "Jordan! Watch This!" → The Video → The Knot in the Stomach → Pause → Think → Act → Walking to a Trusted Adult), each with a detailed image prompt in a <details> block plus prose narration below. Closes with a "What Jordan Teaches Us" reflection table and a "Related Reading" cross-link section back to Chapters 1 and 2.

4. YAML frontmatter quoting bug — silent rendering failure

What changed: Re-wrapped the title: and description: values in double quotes:

1
2
3
4
---
title: "Chapter 2: What Is a Digital Citizen?"
description: "Learn what it means to be a citizen of the digital world  not just a user  and meet Maka's central habit: pause, think, act."
---

Why: A linter (or a manual edit) had stripped the quotes earlier in the session. The title: value contains an embedded colon (Chapter 2: What Is...), which makes the unquoted YAML ambiguous. mkdocs's frontmatter parser silently fails on the ambiguity and falls back to rendering the entire --- block as page content at the top of the published chapter. Quoting fixes this. The same defensive quoting was applied to description: because it also contains a colon-adjacent phrase.

Long-term fix: A new "YAML Frontmatter Quoting" subsection was added to CLAUDE.md (just above the Quick Checklist) with right/wrong examples and a rule: any frontmatter value containing :, #, &, *, ?, |, >, !, %, @, or backtick must be wrapped in double quotes. A new checklist item — "YAML frontmatter title: and description: are wrapped in double quotes" — was added to the chapter-completion checklist.

5. Story-link promotion: admonitions → primary buttons

What changed: The two !!! note "Read Jordan's Story" admonitions that linked to the mini graphic novel were replaced with mkdocs-material primary buttons:

1
[Read Jordan's Story](../../stories/jordan-one-second-choice/index.md){ .md-button .md-button--primary }

Why: The admonition style visually de-emphasized the story link, which is a major piece of the chapter, not a side note. The primary button makes it impossible to miss. Buttons render via the existing attr_list extension in mkdocs.yml — no new infrastructure required.

6. Removed Jordan narrative prose from the chapter body

What changed: Every Jordan-specific narrative detail was either deleted or generalized to second person. Concretely:

Location Before After
Opening section heading ## Jordan's Choice ## Beyond the Words
Opening section body Four paragraphs of playground scenario (recess, the video, the ask, the knot in the stomach) Two paragraphs that frame the chapter and point to the story button. No playground prose.
"What Is a Digital Citizen?" example "When Jordan stops to ask whether sharing that video is kind…" "When you stop to ask whether sharing something is kind…"
"Three Things" example "When Jordan thinks 'the kid in the video deserves to be treated with respect'…" "When you think 'everyone in our class chat deserves to be treated kindly'…"
"Digital Threshold" example "When Jordan's finger is hovering over the share button…" "When your finger is hovering over a share button…"
Closing section ## Jordan's Choice, One More Time — full retelling of the story climax ## You Can Do This Too — a short universal payoff with no playground details

Why: Once the mini graphic novel exists, the chapter prose should not duplicate the story's narrative work. The story is now the canonical place where the playground scene lives. The chapter teaches the concepts directly in second person, and the two prominent buttons make sure readers find the story whenever they want it. This split also removes a real risk: if the chapter prose and the story drift over time, two versions of "what happened to Jordan" could end up contradicting each other.

7. Emoji audit — added 23, kept 2

What changed: Mid-session, 23 emoji were added throughout the chapter for visual interest: 8 in section headings and 15 in the recap list. After a senior-instructional-design audit grounded in Mayer's Cognitive Theory of Multimedia Learning, 21 of the 23 were removed.

Kept:

  • 🚪 on the heading ## The Digital Threshold — directly signals the doorway metaphor the chapter teaches.
  • ⏸️ inline next to the first definition of Pause, think, act — universally recognized pause symbol, reinforces the action at the moment of first encounter, satisfies signaling and spatial-contiguity principles.

Removed (and why each one failed the test):

Emoji Where Why removed
🌐 "Beyond the Words" heading Decorative; doesn't signal anything specific
🌍 "A New Kind of Place" heading Weakly thematic at best
🙋 "What Is a Digital Citizen?" heading A hand-raise doesn't mean "citizen"
🪑 "Three Things..." heading Anti-pedagogical — spoils the stool metaphor before students read it
🛞 "Four Habits..." heading Steering wheel reads as "driving," not "wagon" — wrong metaphor; also spoils the wagon reveal
💪 "You Can Do This Too" heading Tone mismatch — flexed bicep reads as "be tough," not "pause and choose"
📋 "Quick Recap" heading Pure decoration
15 emoji Recap list (one per term) Split visual attention from the bold-and-define pattern; introduced a competing mnemonic system; created confusable icons (four people-icons for four different people-related concepts)

The rule that came out of this audit:

Use emoji only when they signal a metaphor the chapter teaches. Decorative emoji compete with Maka and the bold-and-define pattern — leave them out.

This rule was added to:

  • CLAUDE.md (project-level), as a new "Emoji Use" subsection above the Quick Checklist
  • The chapter-content-generator skill's Best Practices section (item 14), so the rule travels with the skill across all intelligent textbook projects, not just this one
  • A standalone decision log at logs/when-to-use-emoji.md that captures the full reasoning, the audit table, the framework (Mayer's coherence, signaling, spatial-contiguity, redundancy principles), and references to Mayer's Multimedia Learning, Harp & Mayer's seductive-details study, and Sweller's cognitive load theory

The lesson worth keeping for future chapters: "kids find them friendly" is an engagement argument, not a learning argument. Maka the mascot already does the engagement job by design. Emoji on top dilute Maka's role and add extraneous cognitive load that, per Mayer, measurably reduces retention.


Supporting changes (outside this directory but required by the chapter)

docs/stories/jordan-one-second-choice/index.md (new)

The 8-panel mini graphic novel companion to this chapter. Contains a cover image prompt, eight panel image prompts in <details> blocks, prose narration below each panel, an epilogue table, a call-to-action, and a Related Reading section that links back to Chapters 1 and 2. Image generation (cover + 8 panels at ~Create.039 each, ~Create.35 total) is a separate task; the file currently contains only the markdown structure plus the prompts. As of the end of the session, panel-04, panel-05, and panel-07 have been re-pointed to .jpeg files generated externally, and a "Please generate an image for panel 4" line was added to the panel 4 prompt by the user/linter.

mkdocs.yml

  • New Stories section added between Chapters and Learning Graph, containing the link to the Jordan graphic novel
  • pymdownx.emoji extension enabled (with the Material Twemoji index and SVG generator) so any emoji that do survive the audit render consistently as Twemoji SVGs across all browsers and platforms

CLAUDE.md (project-level)

  • New "Mini Graphic Novel Stories" top-level section describing the two-pass workflow (chapter-content-generator creates a stub with a TODO marker; story-generator skill expands it later) and the working example (jordan-one-second-choice)
  • Brief pointer added to the existing "Story and Scenario Patterns" subsection so chapter authors discover the option
  • New "YAML Frontmatter Quoting" subsection with right/wrong examples and the special-character escape rule
  • New "Emoji Use" subsection with the discipline rule and a pointer to logs/when-to-use-emoji.md
  • New checklist item: "YAML frontmatter title: and description: are wrapped in double quotes"

logs/when-to-use-emoji.md (new)

Standalone instructional-design decision log capturing the framework, the audit, the rule, and the references. See section 7 above for details.

chapter-content-generator skill (~/Documents/ws/claude-skills/skills/chapter-content-generator/SKILL.md)

Best Practices item 14 added: Emoji discipline — same one-line rule as in CLAUDE.md, plus a brief Mayer reference. This change is published to the shared skill repo so all intelligent textbook projects pick it up.


Final state of the chapter (end of session 2026-04-11)

Property Value
File docs/chapters/02-what-is-a-digital-citizen/index.md
Body length ~2,000 words (tightened from ~2,300 after Jordan prose removal)
Concepts covered 15 (all from the dependency-ordered list, all bolded inline at definition)
Maka admonitions 5 (welcome, thinking, tip, warning, celebration) — none back-to-back
Inline diagrams 3 (three-legged stool, digital threshold doorway, four-wheeled wagon)
Inline microsims 1 (pause-think-act decision tool, p5.js, spec only)
Tables 1 (opportunities / responsibilities / rights)
Buttons 2 (primary, both linking to the mini graphic novel)
Emoji 2 (🚪 in one heading, ⏸️ inline at one definition)
Trusted-adult rule placement Plain prose only, never inside a Maka admonition
Companion story docs/stories/jordan-one-second-choice/index.md (8-panel graphic novel, prompts only — image generation pending)

Open follow-ups

  • Image generation for the three inline diagram specs (stool, doorway, wagon). These are AI-image-generator prompts only; the PNG files have not been produced yet.
  • Image generation for the Jordan graphic novel cover and remaining panels. The user has begun generating panels (panels 4, 5, and 7 reference .jpeg files), but the full set is not complete.
  • Pre-publication review by a Grade 5 teacher before classroom use. Reading-level check, scenario relatability check, and a read-aloud test with at least one Grade 5 student.
  • Standards alignment cross-reference — confirm that the 15 concepts in this chapter map cleanly to the ISTE Student Standards mentioned in the chapter prose.

Page Feedback — Leave a comment or reaction below. Requires a GitHub account.