Long Article Illustration Assistant

Problem

Long-form articles often need stable illustration placement, consistent visual style, and prompts that produce usable images. This skill addresses the specific problem of turning article sections into concrete image prompts and deciding where inline illustrations should appear, especially for technical posts, newsletters, and product explanations.

How It Works

It starts by collecting the article text, image style, aspect ratio, and illustration density. Defaults include flat illustration style, a 16:9 ratio, and one image every two to three paragraphs. The workflow then splits content by explicit headings, semantic boundaries, and balanced length, aiming for each image to cover 300-600 words and avoiding forced merges across unrelated themes.

For each selected section, it extracts the most visual elements and converts abstract ideas into concrete scenes. Prompts include lighting, composition, color, ratio, and other generation parameters, and explicitly avoid text in the final image. When characters appear, it asks for consistent character identity. The skill can call configured image-generation APIs such as DALL-E, Midjourney, or Stable Diffusion, or output a prompt list for manual generation.

Limits

It is best for illustration planning, prompt generation, and delivery structure, not article rewriting, fact-checking, or rights review. If the source text lacks clear visual subjects, or if the task requires photography, brand assets, or compliance-specific visuals, human review and extra constraints are needed. Output can be a Markdown document, prompt list, or packaged files.