# Concept Comic
<p align="center">
<strong>English</strong> | <a href="./README.zh-CN.md">简体中文</a>
</p>
**Concept Comic** is an AI skill for turning complex ideas into comic-style explanation images.
## ✨ What Makes It Different
**🎯 Expression-point planning**
Concept Comic plans by expression point, not by topic title. A complex concept is split into the smallest visual steps that can still be read quickly.
**🧭 Five-dimensional comic taxonomy**
The skill does not pick a comic type from a flat genre list. It classifies each request across length/form, function, narrative structure, style shell, and channel before writing the storyboard.
**🧠 Automatic type routing**
Concept Comic can route a request to single-panel concept, comparison explanation, four-panel process, dialogue clarification, cause/process comic, system map, journey/state transition, or short vertical comic.
**🧩 Structure before style**
The workflow turns a concept into a concept card, chooses a visual metaphor, writes a storyboard, and only then generates the comic. This keeps the image explanatory instead of decorative.
**🛡️ Channel, style, and risk aware**
Prompts carry audience, channel, risk level, preferred type, and style shell. High-risk topics stay educational, metaphor limits are checked, and public-sharing outputs can reserve AIGC labeling space.
**📚 Built from proven visual-explanation practice**
The skill combines comics, multimedia learning, dual coding, icon usability, and image-prompt practice into one repeatable workflow.
**🔌 Editor-agnostic generation path**
In Codex, use the host image-generation capability directly. In other editors, configure an image API endpoint and key. The default model target is `gpt-image-2`.
```text
OPENAI_API_KEY required outside native image hosts
OPENAI_BASE_URL optional, default: https://api.openai.com/v1
OPENAI_IMAGE_MODEL optional, default: gpt-image-2
```
For a multi-image explanation, make one image request per planned comic image. Do not ask the image model to create the whole series as one combined poster.
## 🧠 Method Sources
Concept Comic packages practical visual-explanation habits from several public bodies of work:
- **Comics and sequential art**: Scott McCloud's work on comics as communication, especially simplification toward meaning. See [Understanding Comics overview](https://www.ebsco.com/research-starters/literature-and-writing/understanding-comics-invisible-art-scott-mccloud) and the [Cleveland Public Library toolkit](https://cpl.org/wp-content/uploads/comic-discussion-guide-Understanding-Comics-Toolkit-11x17-Booklet.pdf).
- **Multimedia learning**: Richard Mayer's principles support removing irrelevant detail, keeping labels close to visuals, and using cues to guide attention. See the Cambridge Handbook chapter on [coherence, signaling, redundancy, and spatial contiguity](https://www.cambridge.org/core/books/cambridge-handbook-of-multimedia-learning/principles-for-reducing-extraneous-processing-in-multimedia-learning-coherence-signaling-redundancy-spatial-contiguity-and-temporal-contiguity-principles/CD5B7AE1279A9AB81F8EEBB53DBEC86E).
- **Dual coding**: Paivio's dual coding theory motivates pairing short verbal labels with strong visual representation. See [Dual Coding Theory and Education](https://nschwartz.yourweb.csuchico.edu/Clark%20%26%20Paivio.pdf).
- **Icon and label usability**: NN/g guidance supports short labels, recognizable icons, and avoiding unlabeled symbolic UI. See [Icon Usability](https://www.nngroup.com/articles/icon-usability/) and [Yes, Icons Need Text Labels](https://www.nngroup.com/videos/icon-text-labels/).
- **Image-prompt practice**: the skill turns those ideas into prompt rules such as short labels, visible action, metaphor-first composition, and no dense infographic or PPT-style layout.
## 🖼️ Real Generated Examples
### RAG vs Knowledge Graph QA
Four points: evidence retrieval, relationship traversal, failure modes, and hybrid use.
### Inflation As A System
Four points: demand surge, supply shock, money versus goods, and purchasing power.
### Natural Selection
Four points: variation, selection pressure, reproduction, and population change.
## 🚀 Use
```text
Use Concept Comic to explain RAG vs Knowledge Graph QA.
```
Expected behavior:
```text
1. Plan how many comic images are needed.
2. Generate separate comic image files.
3. Return image paths.
```
Output contract:
```text
One point = one comic image.
If one point does not fit clearly, split it.
N planned images = N separate generated image files.
Final output is comic imagery, not a text-only explanation.
```
## 📦 Skill Contents
```text
concept-comic/
├─ SKILL.md # Main skill instructions and trigger metadata.
├─ agents/ # UI metadata for skill lists and quick starts.
│ └─ openai.yaml # Display name, short description, and default prompt.
├─ references/ # Load-on-demand guidance used by the skill workflow.
│ ├─ concept-understanding.md # How to extract the core concept and explanation goal.
│ ├─ concept-types.md # Five-dimensional taxonomy and auto-routing rules.
│ ├─ concept-to-metaphor.md # Rules for turning abstract structure into visual metaphor.
│ ├─ metaphor-bank.md # Reusable metaphor patterns.
│ ├─ comic-grammar.md # Comic framing, characters, action, labels, and pacing.
│ ├─ composition-patterns.md # Visual structures, style shells, and channel-aware layouts.
│ ├─ storyboard-template.md # Storyboard fields for each planned comic image.
│ ├─ prompt-template.md # Prompt fields for taxonomy, channel, risk, and style.
│ ├─ image-generation-workflow.md # Native generation route, count planning, and format handling.
│ ├─ external-image-api.md # API fallback for editors without built-in image generation.
│ ├─ qa-checklist.md # Checks for concept, taxonomy, channel, style, safety, and output count.
│ ├─ anti-patterns.md # Common failures such as dense infographics or merged series.
│ ├─ domain-safety.md # High-risk mode, analogy limits, AIGC labels, and IP/style safety.
│ ├─ response-format.md # Final response shape and image path reporting.
│ └─ visual-style.md # Visual style defaults for readable comic images.
├─ examples/ # Text example briefs for common concept requests.
│ ├─ black-hole.md # Example brief for a science concept.
│ ├─ entropy.md # Example brief for entropy.
│ ├─ immune-system.md # Example brief for an analogy-heavy process.
│ ├─ inflation.md # Example brief for an economic system.
│ ├─ natural-selection.md # Example brief for evolution over generations.
│ ├─ opportunity-cost.md # Example brief for a decision concept.
│ ├─ path-dependence.md # Example brief for historical lock-in.
│ ├─ prisoner-dilemma.md # Example brief for game theory.
│ ├─ rag-vs-kgqa.md # Example brief for an AI architecture comparison.
│ └─ water-cycle.md # Example brief for a physical process.
├─ tests/ # Sample pressure prompts for checking skill behavior.
│ └─ sample-prompts.md # Requests that exercise image count and output rules.
├─ docs/ # Repository-facing documentation pages.
│ ├─ gallery.md # Gallery notes for example outputs.
│ ├─ showcase.md # Showcase notes for generated examples.
│ └─ usage.md # Additional usage notes.
└─ assets/ # Images used by README, docs, and examples.
├─ cover.png # Cover graphic asset.
├─ previews/ # Small preview images for older example topics.
│ ├─ entropy-preview.png # Entropy preview image.
│ └─ opportunity-cost-preview.png # Opportunity cost preview image.
└─ examples/ # Generated comic examples and README grids.
├─ .gitkeep # Keeps the examples directory present before assets exist.
├─ rag-vs-kgqa-grid.png # Four-comic grid preview.
├─ inflation-system-grid.png # Four-comic grid preview.
├─ natural-selection-grid.png # Four-comic grid preview.
├─ rag-vs-kgqa/ # Four independent RAG vs KGQA comic images.
│ ├─ 01-rag-find-evidence.png
│ ├─ 02-kgqa-follow-relations.png
│ ├─ 03-different-failure-modes.png
│ └─ 04-hybrid-map-evidence.png
├─ inflation-system/ # Four independent inflation system comic images.
│ ├─ 01-demand-surge.png
│ ├─ 02-supply-shock.png
│ ├─ 03-money-vs-goods.png
│ └─ 04-purchasing-power.png
└─ natural-selection/ # Four independent natural selection comic images.
├─ 01-variation.png
├─ 02-selection-pressure.png
├─ 03-reproduction.png
└─ 04-population-changes.png
```
## License
MIT