by Seyed Yahya Shirazi
Publication-quality figures plugin. Eight skills + one QA agent. figure-bible: scaffolds and validates the per-project theme.json (palette, typography, text limits, Codex model settings) that every other skill reads. ai-full-figure: gpt-image-2 through the Codex CLI (or the OpenAI Images API) renders single panels or whole multi-panel figures with large verbatim titles and panel letters, panels generated in parallel with reference-image consistency, composed at journal width; a text ladder routes dense labels to the SVG overlay and numerals to plot or vector skills. transparent-icons: flat scientific icons with native alpha through the same shared backend, with an explicit Atlas Cloud path. scientific-figure: svgutils composer with pre-export font-size validation and Inkscape/cairosvg export. svg-primitives: mm-precise SVG builder with auto-fit text, edge-snapped arrows, and layer ordering; svg-figure: hand-authoring conventions and editor handoff. plot-styling: library decision tree with SciencePlots recipes. figure-qa agent: type-dispatching QA with OCR of expected strings, palette compliance against the theme, geometry and font checks, a VLM rubric, and a JSON verdict that drives the generate, QA, fix loop.
CodexClaude Code8 Skills