research-paper-figure-skill-factory

Run `npx skills add c-narcissus/research-paper-figure-skill-factory --all -g -y` in your terminal to install it globally.

It analyzes your paper PDFs and figure references, then generates a specialized, reusable skill for one figure type such as method diagrams or taxonomy charts. You then use that generated skill to design, compare, and render figures for any paper.

It enforces a structured workflow: text candidates first, then a diverse board of six candidate images, followed by a second optimization round before the final prompt. This prevents locking into a weak visual direction too early.

Yes. It renders through ChatGPT Images 2.0 on the web and uses the $imagegen skill first in Codex, with packaged skills also portable to OpenClaw and ClawHub.

Yes. The built-in subtype and style atlases show you visual options with embedded reference images, so you pick directions by looking at examples rather than writing design specs.