nature-figure-skill

---
name: nature-figure
description: >-
  Submission-grade Nature/high-impact journal figure workflow. Use whenever the user asks to create,
  revise, audit, or polish manuscript figures, multi-panel scientific plots, figures4papers-style
  matplotlib plots, chart-family recommendations, or journal-ready SVG/PDF/TIFF outputs, especially
  for Nature-family or other high-impact journals. Route no-code / drag-and-drop / "just show me the
  layout" requests to the FigureForge browser editor; otherwise ask "Python or R?" and stop, then use
  only that backend for drawing, previewing, exporting, and QA. Supports matplotlib/seaborn and
  ggplot2/patchwork/ComplexHeatmap. Not for interactive dashboards or Illustrator/Figma-first infographics.
---

# Nature Figure Making Skill

A guide for producing publication-quality scientific figures as a visual argument, not
as isolated pretty plots. Every figure starts from a claim, an evidence hierarchy, and a
review-risk check before code or aesthetics.

If the user does not yet know which chart family fits the task, start with
`references/figure-selection.md` before moving into backend-specific plotting.

This skill covers two routes to the same destination: the no-code **FigureForge** browser
editor (`figureforge/`, 22 publication-grade templates / 16 palettes / 83 atlas panels) and the
scripted **Python / R** pipeline. Triage the route first — see the routing table below.

The older Python/matplotlib rules in this skill remain valid. The skill now also supports
R, especially `ggplot2 + patchwork + ComplexHeatmap + ggrepel + svglite/cairo_pdf + ragg`.
If the user provides a private plotting template collection, use it only as an internal
adaptation source and do not reveal its path, filenames, or provenance in user-facing output.

Color policy: prefer **unified method families across all panels** over maximal hue separation.
For dense Nature Machine Intelligence-style figure pages, use the low-saturation `NMI pastel`
family described in `references/api.md` and reserve green/red mainly for gains, drops, and other directional cues.

## Three hard rules (never trade these away)

**① Type scale is FINAL PRINT SIZE, in points.**
A 183 mm double-column figure uses 5–7 pt text (axis titles 6–7 pt, panel letters 7–8 pt).
Never ship a 16–24 pt default — that is slide-deck size; once the canvas is scaled to journal
width it crops axis titles, overlaps legends and collides tick labels. Start from the
defaults in `references/api.md` (`apply_publication_style()` → 7 pt / 0.8 pt) and only go
larger when the canvas itself is larger. A rendered reference of the defaults lives in
`evals/visual-baseline/` — if you change the defaults, re-render it and look.

**② Legends and stat blocks never sit on top of data.**