Ian Xiaohei Illustrations

Ian Xiaohei Illustrations is a Codex Skill open-sourced by Ian (GitHub user helloianneo) for illustrating Chinese articles, posts, blogs, Notion docs and how-to content. It reads your text, picks out the "cognitive anchors" worth drawing, such as a core claim, two breaking points, a before-and-after or a handoff path, and writes a shot list saying where each image goes, what it shows, what Xiaohei is doing and which short Chinese labels to add. The default is 4 to 8 images per article. You can stop at the shot list or have it call your agent's built-in image tool to make each one: a 16:9 PNG with a pure white background, thin black hand-drawn lines and a few red, orange and blue handwritten Chinese notes, saved to an assets folder in your workspace. What sets it apart is its recurring character. Xiaohei is a solid black little creature with white dot eyes, thin legs and a blank face, and the Skill requires it to carry the main action of each picture, pulling threads, sorting, building a bridge or getting stuck in a gap; if the image still works without Xiaohei, it counts as a miss. The bundled examples are only there to calibrate line density and white space, and the Skill asks for a fresh metaphor from each new article instead of reusing old compositions.

Who it's for:people writing Chinese articles or knowledge and how-to content who want a light, slightly absurd, recognizable illustration style. The README says it isn't meant for commercial illustration, brand key visuals, complex architecture diagrams or cute cartoons, and it doesn't produce editable vector files. You'll need to be comfortable with an AI coding tool such as Codex.

Before you start: the repo contains only text instructions and reference images, with no install script, and the Skill makes no network calls of its own. The README says to clone the repo and copy the ian-xiaohei-illustrations subfolder into Codex's skills directory. It only documents Codex; whether other Skill-capable agents can use it depends on the agent. Image output relies entirely on your agent's built-in image tool (the Skill names image_gen), so without one you only get the shot list; whether that tool goes online and what it costs is set by its provider. The README warns that AI images can come out with wrong Chinese characters, stray titles or style drift, so check each one, and fewer labels give steadier results. The Skill is written in Chinese. Code and docs are MIT licensed. NOTICE.md says Xiaohei is part of Ian's visual language for Chinese article illustrations, that Ian generated the example images, and that anyone redistributing or adapting the repo should keep the Ian Xiaohei Illustrations name or credit Ian. The repo doesn't say whether using the Xiaohei character commercially carries extra limits, so it's worth asking the author before commercial use.

Based on the project's public documentation. Not tested by AI Inspiration World.