Turning Chinese Articles into Xiaohei-Style Illustrations

Where this fits

You're writing a Chinese technical blog post, Notion doc, or workflow guide. The text explains a process, a metaphor, or a state change, but a wall of paragraphs won't stick. You want a quirky, hand-drawn 16:9 illustration that captures the core idea — not a stock photo, not a dense infographic. This skill turns that need into a repeatable agent workflow: it reads your content, decides where images actually help, and generates them in a consistent visual style.

Why agents benefit

• Reads and reasons about content first: The agent extracts cognitive anchors (turning points, loops, comparisons) instead of blindly inserting images after every paragraph.
• Produces a shot list before generating: If you only ask for placement suggestions, it returns a structured list with theme, structure type, and suggested Chinese labels — no image generation needed.
• Generates one image per call: Uses the built-in image_gen tool per illustration, keeping each image focused and avoiding collage-style outputs.
• Enforces a strict visual IP: The "Xiaohei" character must participate in the core action, not stand as decoration. This keeps illustrations cohesive across a document.
• Iterates with a QA checklist: After generating, it checks against style rules (colors, text, composition) and can refine based on your feedback.

Practical scenarios

Blog post on a new methodology

You're publishing a 3,000-word Chinese article about a 5-step workflow. The agent reads the markdown, identifies the two decision points and the final state, and suggests 4 illustrations: one for the overall flow, one for a common pitfall, one for a before/after comparison, and one for the "aha" moment. You approve the shot list, then it generates each image in the Xiaohei style.

Notion documentation for an internal process

Your team uses Notion for runbooks. You paste a page about incident response. The agent extracts the input-output loop and the escalation path, then creates two clean, hand-drawn diagrams with Chinese annotations — making the process scannable at a glance without turning it into a corporate slide.

WeChat article about a product metaphor

You're explaining a complex product as "a backpack that packs itself." The agent suggests a single hero image where Xiaohei is loading a backpack with labeled items, plus a follow-up image showing the unpacked state. The quirky style makes the metaphor memorable and shareable.

Add it to your agent workflow

Install the skill with:

npx skills add Pluviobyte/rnskill --skill ian-xiaohei-illustrations

Then in your agent prompt, you can say: "Read this article draft, generate a shot list, and create the first illustration." The agent will follow the workflow defined in the skill's README, reading reference files as needed.

Compare before adopting

Before committing, compare this skill with alternatives on:

• Quality signals: Does the output truly match the described style (pure white background, red/orange/blue annotations, Xiaohei IP)? Check example outputs or community posts.
• Maintenance freshness: The repo has 1,425 stars and is actively maintained — but verify the last commit date and whether the reference files are updated.
• Alternatives: Other illustration skills might be more generic (e.g., DALL·E prompts) or more business-oriented (infographics). This one is specifically tuned for Chinese text and a quirky hand-drawn aesthetic.
• Workflow fit: Does your agent support the image_gen tool? If not, you'll need to adapt the skill or use a different approach.

Why it is worth tracking

This skill solves a real pain point: converting dense Chinese technical writing into visual explanations that are both informative and charming. The 1,425-star count and active GitHub repo indicate a growing community. Even if you don't need Xiaohei illustrations today, the skill demonstrates a strong pattern — content-aware image generation with a defined style guide and QA loop. Evaluate it when you next need to illustrate a Chinese article, doc, or post. For a broader view of similar agent skills, see the OpenAgentSkill listing.