# ian-xiaohei-illustrations

> Codex skill that draws white-background, hand-drawn illustrations for Chinese articles

- Repository: https://github.com/helloianneo/ian-xiaohei-illustrations
- Category: Agent skills & plugins
- Stars: 12,758 · Forks: 1,463 · Language: n/a · License: MIT
- Experience: Beginner friendly
- Page: https://ir.adyog.com/repos/helloianneo/ian-xiaohei-illustrations/

Chinese writers who want images for an article often get generic illustrations or slide-like infographics. This skill gives the AI agent a fixed visual style and a method, so each image explains one idea from the text instead of just decorating it.

The agent reads your article, Markdown, Notion content or a single topic. It picks one key idea per image, such as a judgment, a flow or a state. It then invents a low-tech, slightly absurd physical metaphor and sends each image to the image model separately. You get a shot list of 4 to 8 images, then PNG files.

The look is a pure white background, thin black hand-drawn lines, lots of empty space and a few red, orange and blue handwritten Chinese notes. The recurring character is Xiaohei, a solid black figure with white dot eyes and thin legs, who must take part in the main action. The README says the example images only calibrate style and should not be copied.

## What it does

- **Shot list planning**: The agent can first list 4 to 8 images for an article. Each entry gives where it goes, the theme, the core meaning, the structure type, what Xiaohei does and suggested Chinese labels.
- **One idea per image**: Each image gets one structure type, such as workflow, before and after, concept metaphor, map route or mini comic panels.
- **Consistent Xiaohei character**: A black hand-drawn figure works inside each scene. The README's test: if the picture still works without Xiaohei, the character is too decorative.
- **QA checklist**: Each image is checked for white background, empty space, Xiaohei's action, short Chinese labels and no copying of old examples. The README warns that image models can produce wrong characters or extra titles.
- **Editing and single-concept modes**: You can plan without generating, generate one image from one sentence, or ask to edit an image, for example to remove a title.

## What you could build

- Generate 4 to 8 inline illustrations for a Chinese blog post or WeChat article.
- Plan where images should go in a long article before drawing anything.
- Turn an abstract idea like trust or workflow handoff into one visual metaphor.
- Give a knowledge or AI-workflow newsletter a recognizable, repeatable image style.
- Fix a generated image by removing a stray title or wrong text.

## Who it's for

Suits Chinese-language content creators who already use Codex and have access to an image model; it is not for editable vector files, commercial illustration or detailed diagrams.

## Get started

Clone the repo, then copy the inner ian-xiaohei-illustrations folder into your Codex skills directory (create it first with mkdir -p if needed). In Codex, call the skill with $ian-xiaohei-illustrations and paste your article.

```
git clone https://github.com/helloianneo/ian-xiaohei-illustrations.git
```
```
cd ian-xiaohei-illustrations
```
```
cp -R ./ian-xiaohei-illustrations "${CODEX_HOME:-$HOME/.codex}/skills/"
```

Tags: codex-skill, illustration, chinese, image-generation, handdrawn