scientific-figure-making
Skill2 starsChanged 13 days agoMITView on GitHub
Create and refine scientific figures with Matplotlib: manuscript data plots, experimental composites, method overviews and image-led research teasers. Use for final-size typography, uncertainty, panel layout, vector export and visual review. Preserve experimental results; use paper-colors for palettes.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- Gavin-Tao/claude-skillstheir own version, first seen Nov 2, 2026
- wangyi0403/dl-research-workflowtheir own version, first seen Oct 13, 2026
- guoyingwei6/scientific-research-skillstheir own version, first seen Jun 16, 2026
- ChenLiu-1996/figures4paperstheir own version, first seen Apr 19, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- ChenLiu-1996/figures4papers★ 12ksame name
- wangyi0403/dl-research-workflow★ 4same name
- Gavin-Tao/claude-skills★ 3same name
- guoyingwei6/scientific-research-skills★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository2 other files it gives its agents.
- paper-colorsSkill
- prismaticaSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 26
Use scripts/paper_layout.py for final dimensions, typography, inspect and PDF/SVG/PNG export.
- Distinctive: line 58
Read only the relevant branch.
- Distinctive: line 68
Do not fake denoising by adding noise to final outputs, invent sequence timing, or interpret attention/projection geometry as causal …
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Scientific figure making
For the combined palette + composition workflow, use the sibling Prismatica entrypoint. This module remains independently usable.
Workflow
- Identify the scientific claim, available data/assets and output venue. For manuscript experiments choose compact data plots, not an editorial poster.
- Preserve values, units, pairing, sample identities and missing observations. Define the experimental unit and whether uncertainty is SD, IQR, SEM or CI. Synthetic examples are fixtures, never reportable research results.
- Choose the relevant reference below. For generic manuscripts start at 85 mm single-column / 178–180 mm double-column and 7–8 pt text, then honor the actual venue template. Nature settings are a separate optional profile.
- Use
scripts/paper_layout.pyfor final dimensions, typography,inspectand PDF/SVG/PNG export. Apply an existing palette; siblingpaper-colorscan generate compatible themes. Never shrink text simply to fit more panels. - Render and inspect the preview at the final insertion size. Check clipping,
labels, legends, line/marker distinguishability and linked image crops.
inspectcatches some text problems, not all collisions or scientific errors. - Deliver the figure, reproducible code, input provenance and statistical definitions. Do not invent missing results, significance or model outputs.
Choose a mode
| Need | Read |
| Manuscript CS experiments: training, alignment, reconstruction | CS paper data |
| Factorial ablation, modality conflict, sampler cost | CS paper comparisons |
| General layout, editable export, final-size review | Layout |
| Atmospheric warm–cool composition, focal detail and asymmetric balance | Landscape aesthetic reference |
| Architectural framing, material contrast and editorial hierarchy | Architectural aesthetic reference |
| Generative method/teaser aesthetics: state, light, point samples | Gamma, SKALAR, Pointillisme studies |
| Nature production profile and source distinctions | Nature |
| Biological data: OncoPrint, spatial omics, annotated trees, genomic Circos, Sashimi | Bioinformatics |
| Set overlaps, flower/petal summaries, Venn/Euler, UpSet, Jaccard | Set overlap |
| Statistical plot families and complex compositions | Gallery catalog |
| CS model analysis and diagnostic examples | Analysis, diagnostics |
| Method overview with scope, mechanisms and evidence | Trinity overview |
| Benchmark-wide iterations, radial bars, correlations | Trinity data reference |
| Image-first CS teasers with ordered results | CS editorial, sequences |
| Quantitative multimodal and diffusion plates | Computed retrieval, conflict and sampling studies |
| Multimodal / diffusion published reference studies | Multimodal, recent studies |
| Curated visual examples with reproducible arrays | examples/showcase_gallery.py and examples/advanced_showcase.py (multi-track, representation and flow compositions); gallery-sized masters, resize typography for a venue |
| Full example commands and specialized chart references | Gallery catalog |
Read only the relevant branch. Scientific data plots take priority over teaser styling when the user asks for an experimental figure.
Provenance boundaries
- Original synthetic templates, actual local demo-model results, published-data redraws and original-paper excerpts are distinct output types. Label them.
- Paper reference tools accept local source PDFs; do not distribute original paper images under this repository's MIT license. Preserve authorship and source identifiers. “Ours” inside a source panel means its paper's authors.
- Do not fake denoising by adding noise to final outputs, invent sequence timing, or interpret attention/projection geometry as causal evidence.
- Retain split denominators, negative outcomes, retrospective Oracle status and per-sample matching. Use a paired or clustered resampling unit when warranted.
Minimal use
import matplotlib.pyplot as plt
import paper_layout as layout # copy from this skill's scripts directory
layout.apply(font_size=8)
fig, ax = plt.subplots(figsize=layout.figure_size("single"), layout="constrained")
ax.plot(steps, accuracy, marker="o", label="Method A") # user's actual arrays
ax.set(xlabel="Training step", ylabel="Accuracy (%)")
ax.legend()
print(layout.inspect(fig))
layout.save(fig, "figures/accuracy")From the repository root, a synthetic fixture smoke run is:
uv run --extra science python skills/scientific-figure-making/examples/cs_paper_data_gallery.py --out /tmp/paper-datasave preserves the requested canvas dimensions; avoid subsequent resizing that
silently changes text size. SVG text remains editable; receiving systems need
its fonts. The full package includes optional sibling paper-colors so the
historical galleries and tests retain their expected relative imports.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

