atomk-hermes-skills/skills/creative/baoyu-visual-content/SKILL.md

name, description, version, author, license, platforms, metadata

name	description	version	author	license	platforms	metadata
baoyu-visual-content	Visual content generation by 宝玉: article illustrations (Type×Style×Palette), knowledge comics, and infographics (21 layouts × 21 styles).	2.0.0	宝玉 (JimLiu) / Hermes Agent	MIT	linux macos windows	hermes tags homepage article-illustration comic infographic creative image-generation baoyu https://github.com/JimLiu/baoyu-skills

Baoyu Visual Content Generation

Three visual content modes from the baoyu-skills project, all built on the same image_generate workflow: analyze → confirm settings → generate prompts → generate images → download & finalize.

Shared Workflow (All Three Modes)

Core image_generate Rules

• image_generate accepts only prompt (string) and aspect_ratio (landscape|portrait|square)
• It returns a URL — always download via curl -fsSL -o <abspath> "<url>"
• Always use absolute paths for download — never rely on CWD persistence across batches
• Never write model names into prompts expecting routing — the backend is user-configured

Shared Principles

• Preserve source data faithfully — no paraphrasing, no summarizing of statistics
• Strip secrets — scan for API keys/tokens/credentials before writing output
• Prompt files are mandatory — every image must have a saved prompt file before generation
• Backup before overwrite — existing files get -backup-YYYYMMDD-HHMMSS suffix
• Reference images: use vision_analyze to extract traits as text; image_generate doesn't accept image inputs

Language Handling

1. User-specified language (explicit option)
2. User's conversation language
3. Source content language Use the input language for ALL interactions including prompts.

---

Mode 1: Article Illustrator

Illustrate articles with Type × Style × Palette consistency.

Dimension	Controls	Examples
Type	Info structure	infographic, scene, flowchart, comparison, framework, timeline
Style	Rendering	notion, warm, minimal, blueprint, watercolor, elegant + 18 more
Palette	Color scheme	macaron, warm, neon — overrides style defaults

Presets combine type+style+palette in one shot (e.g., edu-visual).

Workflow

1. Detect reference images (if provided) → vision_analyze
2. Analyze content → {output-dir}/analysis.md
3. Confirm settings (clarify tool, one question at a time)
4. Generate outline → {output-dir}/outline.md
5. Generate prompts (BLOCKING — every prompt saved before any image) → {output-dir}/prompts/
6. Generate images → download to {output-dir}/
7. Finalize — insert ![description](relative-path) after corresponding paragraphs

Output Directory

• Article file path → {article-dir}/imgs/
• Pasted content → illustrations/{topic-slug}/

See references/article-illustrator/ for full workflow, style gallery, palette options, and prompt construction templates.

---

Mode 2: Knowledge Comics

Create educational/biography/tutorial comics with flexible art × tone combinations.

Visual Dimensions

Option	Values
Art	ligne-claire (default), manga, realistic, ink-brush, chalk, minimalist
Tone	neutral, warm, dramatic, romantic, energetic, vintage, action
Layout	standard, cinematic, dense, splash, mixed, webtoon, four-panel
Aspect	3:4 (portrait default), 4:3, 16:9

Presets

Preset	Equivalent	Hook
ohmsha	manga + neutral	Visual metaphors, gadget reveals
wuxia	ink-brush + action	Qi effects, combat visuals
shoujo	manga + romantic	Decorative, eye details
concept-story	manga + warm	Symbol system, growth arc
four-panel	minimalist + neutral + four-panel	起承转合 structure

Workflow

1. Setup & Analyze content → analysis.md
2. Confirm style & options (clarify tool, one question at a time; handle timeout per-question)
3. Generate storyboard + characters → storyboard.md, characters/characters.md
4. Review outline (if requested)
5. Generate prompts (character descriptions embedded inline) → prompts/
6. Review prompts (if requested)
7. Generate images: character sheet (7.1) then pages (7.2)
8. Completion report

Key Rules

• Character consistency: text descriptions from characters/characters.md embedded in every page prompt
• Character PNG sheet is a human-facing review artifact, NOT an input to image_generate
• Use absolute paths for curl -o (silent footgun: batch CWD drift)

See references/comic/ for full art/tone/preset definitions, layout guides, and partial workflow options.

---

Mode 3: Infographics

Two dimensions: layout (information structure) × style (visual aesthetics). Freely combine any layout with any style.

21 Layouts

linear-progression, binary-comparison, comparison-matrix, hierarchical-layers, tree-branching, hub-spoke, structural-breakdown, bento-grid (default), iceberg, bridge, funnel, isometric-map, dashboard, periodic-table, comic-strip, story-mountain, jigsaw, venn-diagram, winding-roadmap, circular-flow, dense-modules

21 Styles

craft-handmade (default), claymation, kawaii, storybook-watercolor, chalkboard, cyberpunk-neon, bold-graphic, aged-academia, corporate-memphis, technical-schematic, origami, pixel-art, ui-wireframe, subway-map, ikea-manual, knolling, lego-brick, pop-laboratory, morandi-journal, retro-pop-grid, hand-drawn-edu

Keyword Shortcuts

• 高密度信息大图 / high-density-info → dense-modules layout
• 信息图 / infographic → bento-grid layout

Workflow

1. Analyze content → analysis.md (backup existing)
2. Generate structured content → structured-content.md
3. Recommend 3-5 layout×style combos
4. Confirm options (clarify tool)
5. Generate prompt → prompts/infographic.md (load layout+style definitions from references)
6. Generate image
7. Output summary

See references/infographic/ for all layout/style definitions, analysis framework, and structured-content templates.