An OpenAI Codex skill that guides an AI agent to extract cognitive anchors from Chinese-language articles and produce 16:9 hand-drawn 'Xiaohei' style illustrations with sparse Chinese annotations. It targets creators who want memorable visual metaphors rather than generic photos or standard infographics.
Project overview
An OpenAI Codex skill that guides an AI agent to extract cognitive anchors from Chinese-language articles and produce 16:9 hand-drawn 'Xiaohei' style illustrations with sparse Chinese annotations. It targets creators who want memorable visual metaphors rather than generic photos or standard infographics.
Project type
Image & Vision
Use cases
Documents & Office · Learning & Education
Deployment
Refer to project documentation
License
MIT
Best for
Creators and general users who need hand-drawn style illustrations of cognitive actions or metaphors extracted from Chinese articles or Notion content.
Key capabilities
Analyzes Chinese articles, posts, blogs, or Notion content to extract key judgments, processes, structures, states, or metaphors suitable for visualization.
Generates a shot list of 4-8 illustrations per article detailing theme, core meaning, structure type, character action, and suggested Chinese annotations.
Prompts the image model to produce 16:9 horizontal illustrations using a pure white background, black hand-drawn line art, and sparse red, orange, and blue Chinese annotations.
Edits generated illustrations to remove incorrect text or unwanted titles while keeping other content unchanged.
Limitations and risks
The skill does not output editable vector files, SVGs, HTML, or Canvas files.
The skill does not output PPTX, PDF, or Keynote files.
AI image models may produce typos, hallucinated labels, style drift, or unwanted titles in the generated images. The editing capability can help remove incorrect text.
Getting started
Clone the repository, copy the 'ian-xiaohei-illustrations' directory to the Codex skills folder, and invoke the skill via the $ian-xiaohei-illustrations prompt. No coding is required to get started.