What problem does it solve? Researchers often produce figures that mislead readers, fail accessibility checks, or cannot be reproduced, weakening otherwise sound manuscripts. This Skill provides standards for choosing the right chart, encoding data honestly, and generating figures with reproducible code. ## Core Features & Use Cases - Chart Selection & Honest Encoding: Match chart types to data (distributions, comparisons, trends, networks) while enforcing zero-baseline bars, uncertainty bands, and labeled axes with units. - Accessibility & Format Standards: Apply colour-blind-safe palettes (Okabe-Ito, viridis), redundant encoding, and vector (PDF/SVG) or 300-dpi raster output at print scale. - Reproducible Figure Code: Generate matplotlib/seaborn, ggplot2, and plotly scripts plus TikZ, Mermaid, and Graphviz diagrams that regenerate exactly from in-repo data. - Use Case: When drafting a results section for a construction safety study, ask for a violin plot with error bars and colour-blind-safe styling, and receive a ready-to-run matplotlib script exporting a vector PDF. ## Quick Start Ask the assistant to design a publication-quality figure for your results data and generate the reproducible matplotlib or ggplot2 code for it.
