What is ian-xiaohei-illustrations?

This project is a skill for AI coding tools, specifically Codex, that generates hand-drawn illustration images to accompany Chinese articles and blog posts. The goal is not to decorate a page with a generic image, but to identify a key idea, process, or metaphor within the article and draw it as a standalone visual that makes that idea more concrete and memorable. Every illustration follows the same visual style: a pure white background, black hand-drawn line art with thin slightly uneven strokes, generous empty space around the main subject, and a small amount of Chinese handwritten annotations in red, orange, or blue. The central character in most images is called Xiaohei, a simple black solid figure with white dot eyes and thin legs. Xiaohei is meant to be an active participant in whatever the diagram depicts, not a decoration standing to the side. The README describes the aesthetic as strange but clean, and explicitly rules out cute, childish, or cartoon-style results. The typical workflow starts by reading the article and identifying four to eight moments worth illustrating. For each one the agent produces a shot list entry describing where in the article it belongs, what the core idea is, and what Xiaohei should be doing. Then it calls an image generation model once per image, checks the result against a quality checklist covering white background, correct proportions, readable Chinese text, and Xiaohei's involvement in the action, and saves the final PNGs to a folder named after the article. The README notes that Chinese text inside generated images should be kept very short for reliability, and that AI image models may occasionally produce typos, misplaced labels, or style drift that requires a manual check before publishing. The skill is installed by copying a subfolder into the directory where Codex looks for skills. The README is written in Chinese.