helloianneo/ian-xiaohei-scenes

This repository is a skill for the Codex AI agent tool, designed to help people who write Chinese-language articles generate matching illustration images. The core idea is to take something described in an article, such as a stressful situation at work or a turning point in a project, and convert it into a small visual scene featuring a character called Xiaohei alongside a real physical object, a physical action, and short Chinese text labels. The goal is to produce an image where a reader immediately recognizes their own experience. The README is written in Chinese. The skill is called Xiaohei Scenes 2.0, and it differs from an earlier version (Xiaohei Illustrations 1.0) in that it focuses on depicting realistic situations and emotions rather than diagrams or process breakdowns. Where the previous version was suited to explaining methods and structures, this version is suited to capturing workplace pressure, AI-era anxieties, product journeys, and personal experiences. There are two modes. The standard mode produces a 16:9 landscape image with a plain white background, one main physical object, one physical action performed by Xiaohei, and two to four short Chinese label tags with minimal color accents. The easter egg long-scroll mode produces an extra-wide horizontal story image for tracing a timeline of personal or project history, with a hand-drawn curved path, five to eight object nodes, and Xiaohei participating in each node. Installing the skill means cloning the repository and copying the inner subfolder into the Codex skills directory. Once installed, you invoke it inside Codex by referencing the skill name and describing your article or situation. The skill guides the AI through reading the content, identifying the core conflict or situation, selecting a reference template image from seven provided examples, choosing a real object and action, generating the image, and then checking it against a quality checklist. The repository includes seven example images, prompt templates, and reference documents covering the character design, object selection patterns, story extraction process, and quality criteria.