# Nature Figure Master > Create publication-grade scientific figures and drive desktop Origin/OriginPro from Codex or Claude Code. Use for SCI, Nature/Science/Cell/IEEE/Elsevier, and thesis figures; automatic data inspection and chart recommendation; journal-style inference from PDFs, websites, or DOIs; Origin template automation; editable OPJU projects; batch export; redraw, beautification, export, or critique from CSV/TSV/TXT/XLSX/XLS, Python/Matlab/Origin outputs, screenshots, diagrams, or draft manuscript figures. - Skill: `zhaolong666520/nature-figure-master` (Agent Skill, multi-file: 7 files) - Install (CLI): `npx skillmds@latest add zhaolong666520/nature-figure-master` - Raw SKILL.md: https://api.skillmd.com/api/skills/zhaolong666520/nature-figure-master/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Data & Analytics - Author: Zhaolong666520 (https://skillmd.com/u/zhaolong666520) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/zhaolong666520/nature-figure-master --- # Nature Figure Master ## Workflow 1. Identify the figure job: data plot, mechanism diagram, workflow schematic, experimental pipeline, multi-panel figure, or thesis visual. 2. Ask only for missing essentials that cannot be inferred: target journal/template, figure size, panel labels, input data, and preferred language. 3. Choose the production route: - For user-dropped data files, inspect extension and table shape first: CSV/TSV/TXT, JSON records, XLSX/XLS workbook, or copied table text. - Recommend chart types from data semantics before drawing: trend line for ordered numeric X, scatter for numeric correlations, bar for categorical group comparison, heatmap for numeric matrices. - Infer journal style from a PDF filename/metadata/text, DOI, domain, or journal name when supplied. Use conservative rules and tell the user when confidence is low. - Import plotting briefs, advisor notes, lab prompt templates, or journal format guides and convert requirements such as "double column", "300 dpi", "TIFF", "Nature", "bar chart", or "thesis figure" into figure settings. - For tutorial-style requests such as Nature multi-group bar charts, offer one-click recipes before manual tuning: Nature grouped bar, SCI trend line, correlation scatter, thesis heatmap, or auto-best. - Use `scripts/style_recipe.py` to generate a journal style JSON when a plotting script or app needs deterministic style settings. - When the user asks to control Origin, first run `python scripts/origin_driver.py doctor`, then inspect the data with `python scripts/origin_driver.py inspect `. - For external Origin automation, read `references/origin-automation.md`, run a `--dry-run` command, review the resolved mappings, and only then run without `--dry-run`. - Require Windows, a licensed local Origin/OriginPro installation, Windows Python, and the official `originpro` package for real Origin execution. On other systems, explain the limitation and use a non-Origin plotting route only with the user's knowledge. - Read `references/figure-standards.md` for publication visual rules when designing or reviewing figures. - Read `references/workflows.md` for multi-panel SCI figure assembly and thesis workflows. - For an app or UI, prefer one smart import center for data, journal/style source, and plotting brief; keep the figure preview centered; put deeper controls into clear tabs rather than a crowded toolbar. 4. Produce editable outputs first: SVG/PDF for vector graphics, PPTX/AI-compatible SVG for diagrams, PNG/TIFF only as final raster exports. 5. Verify the figure at manuscript size: label readability, line weight, color contrast, legend placement, panel alignment, and caption consistency. 6. For Origin jobs, verify every requested export is non-empty and preserve the editable `.opju` project. ## Design Rules - Prefer clean vector marks, restrained color, direct labels, and generous white space. - Use colorblind-safe palettes unless the user provides a journal or lab palette. - When the user asks for language switching or international use, keep the program UI multilingual while leaving figure title/axis text editable by the user. - When the user asks for color choice, expose direct color controls and preserve separate palettes per journal preset. - When building a figure app, show palette presets with actual swatches, support eyedropper/color picker where available, and keep chart colors diverse but colorblind-aware. - Keep the program surface clean: compact import, centered publication preview, tabbed advanced settings, visible export controls, and minimal top-level chrome. - Keep all visible app labels wired to the language switch, including import cards, tab names, plot controls, export labels, and publisher presets. - Provide Origin-style fine controls when practical: legend labels and placement, line width, marker size, frame visibility, border style, grid, smoothing, and per-series colors. - Keep axis titles explicit with units; avoid unexplained abbreviations. - Avoid decorative gradients, shadows, 3D chart effects, and dense gridlines for SCI plots. - Use consistent panel letters: bold lowercase or uppercase at the same x/y position across panels. - Make statistical annotations minimal and reproducible: test name, n, error definition, and p-value style must be stated in caption or methods. - For thesis figures, prioritize readability after Word/PDF compression: slightly larger fonts and stronger line weights are acceptable. - Match Origin-style export expectations when asked: direct SVG/PNG/JPEG/BMP/PDF when available, and provide an editable SVG/PDF master for legacy or proprietary targets such as EMF, WMF, AI, CGM, DXF, PSD, PCX, TGA, EPS, GIF, or TIFF when browser-native generation is not reliable. ## One-Click Recipes - Nature multi-group bar: grouped bars, Nature palette, no heavy grid, outer frame, legend enabled, vertical group separators, concise axis labels. - SCI trend line: line plus markers, journal palette, minimal grid, direct comparison across multiple Y columns. - Correlation scatter: scatter marks for numeric columns, light grid, legend retained, clean correlation-ready canvas. - Thesis heatmap: numeric matrix overview, larger serif font, strong readability for Word/PDF thesis export. - Auto-best: infer chart type from table shape, keep current journal style, and apply low-risk polish defaults. - Graph Publisher-style presets: journal single-column, journal double-column, thesis figure, and presentation slide. Apply size, style, and export defaults together. ## Output Checklist Before finalizing, confirm: - Figure opens correctly in browser, Word, PowerPoint, LaTeX, or the target editing tool. - SVG/PDF contains text and paths cleanly; no accidental bitmap-only export unless requested. - Fonts are common or embedded/outlined when necessary. - Colors remain distinguishable in grayscale. - The final filename includes figure number or short semantic name, such as `fig2-dose-response.svg`. ## Script Use Generate style settings: ```bash python scripts/style_recipe.py --preset nature --output style.json ``` Supported presets: `nature`, `science`, `cell`, `thesis`, `presentation`. Inspect an input without opening Origin: ```bash python scripts/origin_driver.py inspect data.csv ``` Preview the resolved Origin job: ```bash python scripts/origin_driver.py plot data.csv --chart auto --x 0 --y 1 2 --preset nature --formats png,pdf,svg --dry-run ``` Run the same command without `--dry-run` to launch the official Origin automation bridge, export the figures, and save an OPJU project.