Paper Comic

This turns academic papers into visual method breakdowns, focusing entirely on how the thing actually works rather than background fluff. It reads your PDF or arXiv link, recommends a set of diagrams (cover, overview, mechanism details, key results), then waits for you to confirm language, style, and scope before generating anything. You get either sketchnote style (warm, hand-drawn, good for explanations) or paper-figure style (clean vectors, good for READMEs). The core philosophy is tight: one diagram should make one mechanism clear to someone who hasn't read the paper, or you split it into two. It won't waste pages on related work or the fifth ablation study. Works with whatever image generation you have installed, no hardcoded APIs.

Install to Claude Code

npx -y skills add zsyggg/paper-craft-skills --skill paper-comic --agent claude-code