zine-deck-maker

KO | EN

gitlite — search

#typescript #ai-agents #ai #dsh-plugin #deepseek-harness #open-source #claude-code #codex #cli #developer-tools #react #windows

zine-deck-maker

★ 28 Open GitHub ↗

Portable Agent Skill for generating K-pop apocalypse ransom-zine PDF decks from topics.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Zine Deck Maker

A portable Agent Skill package for turning a topic into a 7-8 page PDF or slide deck using a loud K-pop apocalypse ransom-zine visual style.

It is not limited to Codex. The core workflow is platform-agnostic, and adapters are included for Codex, Claude Code, Cursor, ChatGPT, and generic agents.

Important: the example outputs were produced in an environment with GPT-image support. GPT-image is not required. You can use any image generation model or manual design tool by following the provider slot in `core/image-generation.md`.

## Preview

Style reference:

![K-pop apocalypse ransom zine 16:9 reference](https://raw.githubusercontent.com/VigoZhao/zine-deck-maker/main/assets/01-subway-idol-signal-16x9.png)

Contact sheet:

![K-pop apocalypse ransom zine contact sheet](https://raw.githubusercontent.com/VigoZhao/zine-deck-maker/main/assets/style-contact-sheet.png)

Example output preview:

![AI music label content-first example](https://raw.githubusercontent.com/VigoZhao/zine-deck-maker/main/examples/ai-music-label-content-first-preview.jpg)

## What It Does

- Takes a topic, product, idea, trend, story, or campaign.
- Plans 7-8 pages as a `PPT 内容制作大师`.
- Writes short deck-ready Chinese or English copy.
- Generates a PDF-first zine deck with torn paper, crumpled black texture, ransom typography, lime/blue/red accents, and bottom masthead strips.
- Uses content-relevant cutouts on every page: person, product, object, device, interface, receipt, package, scene fragment, map, machine, shelf, or prop.
- Lets users bring their own image model: GPT-image, Midjourney, Flux, SDXL, Ideogram, Firefly, internal models, or manual design tools.
- Falls back to page-by-page image prompts when the current platform cannot generate images directly.
- Avoids unrelated repeated people, clean corporate templates, watermarks, QR codes, copied logos, and exact source names.

## Repository Structure

```text
zine-deck-maker/
├── SKILL.md                         # Codex-compatible root skill
├── core/
│   ├── skill.yaml                   # platform-neutral metadata
│   ├── instructions.md              # platform-neutral workflow
│   ├── image-generation.md          # bring-your-own image model slot
│   ├── style-spec.json              # machine-readable style rules
│   ├── style-guide.md               # human-readable style guide
│   └── prompt-template.md           # reusable prompt templates
├── adapters/
│   ├── codex/SKILL.md
│   ├── claude-code/agent.md
│   ├── claude-code/command.md
│   ├── cursor/rules.md
│   ├── chatgpt/custom-instructions.md
│   └── generic/system-prompt.md
├── assets/
│   ├── style-contact-sheet.png
│   └── reference images
├── examples/
│   └── ai-music-label-content-first-preview.jpg
└── references/                      # backward-compatible copy of style refs
```

## Use With Codex

Clone into your Codex skills directory:

```bash
git clone https://github.com/YOUR_USERNAME/zine-deck-maker.git ~/.codex/skills/zine-deck-maker
```

Restart Codex or start a new session, then use:

```text
使用 $zine-deck-maker，主题：AI 音乐厂牌，做 7 页，16:9，中文文案，输出 PDF。
```

## Use With Claude Code

Use the included adapter as a project or user-level subagent:

```bash
mkdir -p .claude/agents .claude/commands
cp adapters/claude-code/agent.md .claude/agents/zine-deck-maker.md
cp adapters/claude-code/command.md .claude/commands/zine-deck-maker.md
```

Then invoke:

```text
Use the zine-deck-maker subagent. Topic: AI 音乐厂牌. Make 7 pages, 16:9, Chinese copy, output PDF.
```

Or:

```text
/zine-deck-maker 主题：AI 音乐厂牌，做 7 页，16:9，中文，输出 PDF
```

## Use With Cursor

Copy or reference:

```text
adapters/cursor/rules.md
core/instructions.md
core/style-spec.json
core/style-guide.md
```

Then ask Cursor:

```text
Use zine-deck-maker. Topic: AI 音乐厂牌. Make 7 pages, 16:9, Chinese, output PDF.
```

## Use With ChatGPT Or Other Agents

Use:

```text
adapters/chatgpt/custom-instructions.md
```

or:

```text
adapters/generic/system-prompt.md
```

Attach or reference the `core/` files and `assets/` images as the skill's source of truth.

## Bring Your Own Image Model

This skill separates the deck workflow from the image generator.

If your platform has native image generation, use that. If it does not, ask the agent to export one visual prompt per page, then paste those prompts into your chosen image model or design tool.

Supported routes include:

- GPT-image
- Midjourney
- Flux
- Stable Diffusion / SDXL
- Ideogram
- Adobe Firefly
- internal image models
- manual Photoshop, Figma, Canva, or other design workflows

Use this field in your request when you want to specify a provider:

```text
Image provider: Midjourney
```

or:

```text
Image provider: my internal image model
```

If no image provider is available, use:

```text
Use zine-deck-maker. Topic: AI 音乐厂牌. Make 7 pages, 16:9, Chinese. Output page-by-page image prompts and deck layout instructions.
```

## Universal Request Template

```text
Use zine-deck-maker.
Topic: <topic>
Pages: <7 or 8>
Aspect ratio: <9:16 or 16:9>
Language: <Chinese or English>
Output: <PDF, PPTX, or both>
Content depth: <visual-first or content-first>
Image provider: <GPT-image, Midjourney, Flux, SDXL, Ideogram, Firefly, internal model, manual, or unspecified>
```

## Default Behavior

- Output: PDF
- Page count: 7-8 pages
- Copy depth: light planning
- Page content: headline, short caption, sticker words, visual direction
- Content-first option: core claim plus 2-3 key points per page
- Style source: `core/style-spec.json`
- Image generation: provider-agnostic; see `core/image-generation.md`

## License

MIT

zine-deck-maker

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README