skills.rest
Light0305
Light0305COMMUNITY
Explore all skills by this author →
GitHub

light-figure

Generate publication-grade scientific figures programmatically with visual honesty gates and journal specs.

572|72|Updated Jun 7, 2026
Search Tags:#publication#ggplot2#colorblind-safe#Matplotlib#scientific figures#reproducible plots#visual honesty
One-click install
npx skills add https://github.com/Light0305/Light-skills --skill light-figure
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: light-figure
Source: https://github.com/Light0305/Light-skills/tree/main/skills/light-figure
Command: npx skills add https://github.com/Light0305/Light-skills --skill light-figure

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires matplotlib, numpy, pandas, seaborn, and includes scripts (resource) and assets (resource) components.

What problem does it solve?

Scientific figures often mislead readers through truncated y-axes, dual-axis pseudo-correlations, jet/rainbow colormaps, missing error bars, or by presenting non-significant results as main figures. This Skill turns figure creation into a gated, reproducible workflow where every figure is bound to a claim, meets journal column-width and font specifications, uses colorblind-safe palettes, and passes deterministic visual-honesty checks before delivery.

Core Features & Use Cases

  • Visual Honesty Gate: Static linting of plotting code detects y-axis truncation, twin axes, jet/rainbow colormaps, missing error bars, and 3D distortion as critical or warning findings.
  • Publication-Grade Export: Built-in journal specs (Nature, Science, Cell, PLOS, IEEE, Elsevier, MDPI) enforce physical column widths in mm, minimum font sizes, DPI, and vector/raster output formats.
  • Claim-Bound Figure Planning: Plan cards bind each figure to a claim with evidence strength, audit display-item budgets against venue caps, and detect redundant panels.
  • Python + R Dual Path: Detects Rscript/ggplot2 availability and renders statistical figures (forest plots, boxplots, facets) via ggplot2, with honest matplotlib fallback.
  • Use Case: After finishing result analysis, plan a figure set for a Nature submission, render colorblind-safe multi-panel figures with error bars and significance markers, then run the honesty gate and render-then-look visual QA before submission.

Quick Start

Ask the AI to plan and generate publication-grade figures for your paper claims, then run the visual honesty gate on the plotting code before exporting at the target journal's column width.

Frequently Asked Questions about light-figure

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make publication-quality figures with matplotlib?▼

Set the figure to the target journal's physical column width in millimeters, use vector PDF output at 600 DPI, and verify scaled font sizes meet the journal minimum. The figure_export module provides save_for_journal and check_figure_size functions for journals like Nature, Science, and PLOS.

What color palette should I use for colorblind-safe scientific figures?▼

Use the Okabe-Ito palette for up to eight discrete categories and viridis or cividis for continuous data. Avoid jet and rainbow colormaps, which are not perceptually uniform and fail colorblind readers; add redundant encoding like line styles or markers for grayscale printing.

Can I use ggplot2 in R instead of matplotlib for paper figures?▼

Yes, the r_ggplot script detects whether Rscript, ggplot2, and scales are installed and renders forest plots, boxplots, and faceted figures through ggplot2 when available. If R is unavailable, it honestly degrades to matplotlib and marks the output as degraded rather than pretending to produce ggplot figures.

Why is truncating the y-axis considered misleading in bar charts?▼

Truncating the y-axis on bar charts exaggerates small differences because bar length encodes value proportionally. Bars must use a zero baseline; if truncation is justified, it requires a broken-axis mark and an explicit caption note, and the honesty gate flags it as critical for human review.

Does this skill support AI-generated images for paper figures?▼

No, all data figures must be generated programmatically with matplotlib, seaborn, R, or TikZ so they are reproducible and auditable. Generative image models are never used for research data figures, as a permanent non-negotiable rule.

What are the limitations of automated figure honesty checks?▼

Static linting only detects suspicious patterns like nonzero y-limits or twinx usage, not actual misleading intent, since truncation can be legitimate. Final judgment of whether a figure misleads still requires the caption plus human or reviewer evaluation.

Related Skills

Explore More Education & Research Skills→
EvoClawEvoClaw
11

claim-evidence-alignment

Validate manuscript claims against linked figures, tables, experiments, and statistical results.

Official
Advanced
nahisahonahisaho
3

scientific-latex-formatter

Convert Markdown manuscripts into LaTeX documents with journal templates.

Community
Advanced
Light0305Light0305
514

light-figure-planning

Plan publication-ready figures and tables for scientific manuscripts.

Community
Advanced
stellaromicsstellaromics

bio-data-visualization-interactive-visualization

Generate interactive omics visualizations with plotly and bokeh.

Official
Advanced