generate

name	generate
description	Generate rich outputs from ingested sources — flashcards, quizzes, reports, slide decks, mind maps, infographics, data tables, or audio overviews. Use when the user wants to create any output from their documents.
argument-hint	<type> [topic]
compatibility	Requires Bun 1.0+ and Node.js 18+. Audio requires macOS (uses `say` command).
allowed-tools	Bash(bash *) Bash(bun *) Bash(node *) Write Read
effort	high

Generate Output

Generate one of 8 output types from ingested document sources.

Output types

Type	Command	Output files
flashcards	/notebook:generate flashcards [topic]	.json + .md + .html (interactive flip cards)
quiz	/notebook:generate quiz [topic]	.json + .md + .html (50 Qs, live scoring)
report	/notebook:generate report [topic]	.md + .docx
slide-deck	/notebook:generate slide-deck [topic]	.pptx
mind-map	/notebook:generate mind-map [topic]	.mmd + .html (Mermaid viewer)
infographic	/notebook:generate infographic [topic]	.html
data-table	/notebook:generate data-table [topic]	.csv + .json + .md + .html (sortable/filterable)
audio-overview	/notebook:generate audio-overview [topic]	.wav (Kokoro TTS)

Workflow

1. Retrieve context — query ingested sources:

bun "${CLAUDE_PLUGIN_ROOT}/skills/ingest/scripts/query.mjs" "$ARGUMENTS" --top-k 15

2. Build input JSON — see the relevant template in assets/ for the schema:

Type	Template	Input key fields
flashcards	assets/flashcards.json	[{ "front", "back", "tags" }]
quiz	assets/quiz.json	{ "title", "questions": [{ "question", "type", "options", "answer", "explanation" }] }
report	assets/report-docx.json	{ "title", "sections": [{ "heading", "content" }] }
slide-deck	assets/slide-deck.json	{ "title", "slides": [{ "title", "bullets"/"body", "notes" }] }
mind-map	assets/mind-map.mmd	{ "label", "children": [...] } or raw Mermaid
infographic	assets/infographic.json	{ "title", "sections": [{ "heading", "content", "icon", "stat" }] }
data-table	assets/data-table.json	{ "headers": [...], "rows": [[...]] }
audio-overview	assets/audio-overview.json	{ "segments": [{ "text", "voice" }] }

Write input JSON to output/_<type>_input.json.

3. Run the generator:

# Flashcards (zero-dep)
node "${CLAUDE_SKILL_DIR}/scripts/flashcards.mjs" -i <input> -o output --name flashcards

# Quiz (zero-dep)
node "${CLAUDE_SKILL_DIR}/scripts/quiz.mjs" -i <input> -o output --name quiz

# Report (bun auto-installs deps)
bun "${CLAUDE_SKILL_DIR}/scripts/report.mjs" -i <input> -o output --name report --format both

# Slide deck (bun auto-installs deps)
bun "${CLAUDE_SKILL_DIR}/scripts/slide-deck.mjs" -i <input> -o output --name slide-deck

# Mind map (zero-dep)
node "${CLAUDE_SKILL_DIR}/scripts/mind-map.mjs" -i <input> -o output --name mind-map

# Infographic (zero-dep)
node "${CLAUDE_SKILL_DIR}/scripts/infographic.mjs" -i <input> -o output --name infographic

# Data table (bun auto-installs deps)
bun "${CLAUDE_SKILL_DIR}/scripts/data-table.mjs" -i <input> -o output --name data-table

# Audio overview (Kokoro TTS neural voice via uv, falls back to macOS say)
uv run --script "${CLAUDE_SKILL_DIR}/scripts/audio-overview.py" -i <input> -o output --name audio-overview

4. Report output paths. For HTML outputs, tell user: open output/<name>.html

Quality gates

• Flashcards: 15-30 cards, no yes/no questions, tagged by topic
• Quiz: 50 questions, 70% MC / 30% SA, explanations for all
• Report: executive summary, evidence-backed findings, actionable conclusions
• Slide deck: 8-15 content slides, 3-5 bullets each, speaker notes
• Mind map: 3-6 main branches, 2-4 sub-topics each, concise labels
• Infographic: 5-8 sections, stats highlighted, concise
• Data table: specific data points, consistent column formatting
• Audio: 8-15 alternating segments, conversational tone, 2-4 sentences each

Gotchas

• Always query sources first — ground all content in source material
• Slide deck: bullets and body are mutually exclusive per slide
• Audio: uses Kokoro TTS (82M neural model). Best voices: af_heart (female, A-grade), am_fenrir (male, C+). Falls back to macOS say if Kokoro not installed. Strip URLs and special characters.
• Mind map: Mermaid syntax is indentation-sensitive, avoid ()[] in labels
• Report DOCX: supports paragraphs and headings only (no tables/images)
• Quiz MC: answer field is the letter (A/B/C/D), not the full text

Additional resources

• Voice options for audio: references/voices.md

$ARGUMENTS