Cognitive Anchor Sketcher

Most article visuals miss the point. They are either decorative images that say nothing, or dense diagrams that make the reader work harder.

Cognitive Anchor Sketcher starts with the article, not the image prompt. It reads Chinese long-form content, finds the judgment, process, structure, state, turning point, or metaphor worth remembering, and turns that idea into a focused 16:9 hand-drawn editorial illustration.

The problem

Writing a good image prompt is not the hardest part. The harder questions are: What deserves an image? Where should it appear? What action will make the idea visible? And how do you keep several images consistent without turning the article into a slide deck?

This skill handles those editorial decisions before generation begins.

What it does

• Analyzes Chinese articles, blog posts, Notion pages, workflow documents, and methodology content.
• Identifies the strongest cognitive anchors instead of illustrating every paragraph.
• Builds a reviewable shot list with placement, core meaning, composition, character action, and short Chinese annotations.
• Confirms the direction, drawing style, character or IP, image count, and output settings before generation.
• Generates one image at a time, checks the result, and saves approved copies without overwriting existing files.
• Supports two visual presets: minimal-line and emotion-doodle.
• Uses Xiaohei by default, with optional TuoTuo, XingBi, a paired-character mode, or an authorized custom IP.

What you get

A typical article receives a 4–8 image plan. Each image carries one idea, uses a clean white background, black-led hand-drawn linework, generous whitespace, and only a small amount of color where it helps the reader.

You can stop after the cognitive-anchor analysis, continue to a complete shot list, generate only the first image as a test, or proceed through the full approved set.

Why this is different

A normal image prompt starts too late. Cognitive Anchor Sketcher first decides what the image needs to explain and how the selected character should take part in the idea.

The result is not a slide deck pretending to be an illustration. It is a visual that supports the paragraph beside it.

Best for

Writers, product designers, AI builders, Notion users, knowledge-base teams, and anyone publishing Chinese long-form content with abstract ideas.

It is especially useful for product thinking, software workflows, methodology articles, knowledge management, technical writing, and opinion-led content.

Not built for

This is not a PPT, PDF, SVG, data-chart, photorealistic key visual, commercial poster, or silent bulk-generation tool. It also will not copy a custom character when the reference source or permission is unclear.

Compatibility

Analysis, shot-list planning, and prompt writing work in compatible SKILL.md agents. Automatic image generation is designed for Codex CLI with an active ChatGPT login and its built-in image generation tool. No OpenAI API key is required.