What problem does it solve?
This Skill prevents manuscript figures from becoming “pretty but unconvincing” by forcing a clear scientific claim, evidence hierarchy, and journal-grade export/QA workflow before any plotting happens.
Core Features & Use Cases
- Figure contract first: Builds a claim + archetype + panel map (what each panel proves) and enforces reviewer-risk checks before styling or code generation.
- Backend-exclusive plotting: Works with either Python (matplotlib/seaborn) or R (ggplot2 + patchwork + ComplexHeatmap) and prohibits cross-rendering between languages.
- Publication-grade outputs: Produces editable SVG as the primary artifact (plus optional PNG/PDF/TIFF) with strict font/SVG settings to keep text editable.
- Nature/NMI visual discipline: Enforces minimalist spines/legends, cohesive palettes across panels, panel-label conventions, and export bundle completeness (script + source data + QA notes when needed).
Quick Start
Ask the assistant to generate a Nature-style multi-panel figure and include “Backend: Python or R”, the one-sentence core conclusion, the panel map (a/b/c…), and the required output formats (editable SVG plus optional PNG/PDF/TIFF).