scientific-figure

Name: scientific-figure
Author: neuromechanist

37stars7forks

How to install scientific-figure?

neuromechanist/research-skills/scientific-figure

$npx -y skills add neuromechanist/research-skills --skill scientific-figure

Installs into the current project.

Summary

This skill should be used when the user asks to "create a figure", "make a scientific figure", "create a paper figure", "make a figure for my paper", "make a figure for my manuscript", "compose figure panels", "assemble figure panels", "combine figure panels", "make a multi-panel

Files · 1

View on GitHub

SKILL.md

1	# Scientific Figure
2
3	Compose publication-quality multi-panel figures at exact journal dimensions (mm/pt), validate font sizes against journal minima before export, and write PDF/PNG that the journal will accept without resize.
4
5	## Why this skill exists
6
7	The previous react-pdf workflow had two structural failures:
8
9	1. Composition did not respect physical size. Flexbox layout shifted dimensions in subtle ways during render.
10	2. Fonts shrank below readable thresholds. When content overflowed, react-pdf uniformly scaled the entire figure, sometimes pushing axis labels under 4 pt.
11
12	This skill replaces that workflow. Panels are placed at exact mm coordinates, text is preserved as SVG `<text>` elements so font sizes are inspectable before export, and the validator rescales individual panels rather than the whole figure when a font minimum is violated.
13
14	## Pipeline
15
16	```
17	1. Plan 2. Build elements 3. Compose 4. Validate fonts 5. Export
18	(journal (matplotlib/seaborn (svgutils, at (per-element pt vs (Inkscape if
19	size, panel per panel; SVG out; exact mm/pt; journal minimum; present;
20	grid) optional icons) text preserved) rescale panel if cairosvg
21	below minimum) otherwise)
22	```
23
24	Every step uses on-the-fly execution via `uv run --with` so no permanent installs are required for the skill itself.
25
26	## Step 1: Plan the figure
27
28	Before generating anything, fix the following:
29
30	- Target journal — sets the canvas width. See `references/journal-specs.md` for the full table; the four most common are summarized below.
31	- Panel grid — 1x1, 1x2, 2x2, wide-top + sub-panels, or freeform mm coordinates.
32	- Color palette — pick from `references/color-palettes.md` and reuse across all panels.
33
34	### Journal width / font minimum cheat sheet
35
36	\| Journal \| 1 column \| 2 column \| Min body font \| Notes \|
37	\|---\|---\|---\|---\|---\|
38	\| Nature \| 89 mm \| 183 mm \| 5 pt \| 8 pt for panel labels \|
39	\| Science \| 55 mm \| 120 mm \| 6 pt \| Myriad/Helvetica preferred \|
40	\| Cell \| 85 or 112 mm \| 174 mm \| 6 pt \| Max 225 mm tall \|
41	\| PNAS \| 87 mm \| 180 mm \| 6 pt \| No label below 2 mm tall \|
42
43	The validator in step 4 enforces these. Pick the journal early so the validator can warn early.
44
45	## Step 2: Build the elements
46
47	Each panel is generated independently as an SVG. Common sources:
48
49	- Plots: matplotlib/seaborn, saved with `bbox_inches='tight', transparent=True, format='svg'`. See the `[[plot-styling]]` skill for the library decision tree and SciencePlots recipes.
50	- Icons: transparent PNGs from the `[[transparent-icons]]` skill.
51	- Schematics: for new Python-driven work use `[[svg-primitives]]` (auto-fit boxes, validated arrows, layered z-order); for hand-authored SVG patterns see `[[svg-figure]]`.
52
53	Save each element to a working directory (typically `panels/`), then compose them in step 3.
54
55	## Step 3: Compose the figure
56
57	The composer is built around [svgutils](https://svgutils.readthedocs.io/) (MIT license; `uv run --with svgutils`). It places panels at exact mm coordinates and preserves text as inspectable SVG `<text>` elements.
58
59	Two ways to compose: the `Figure` helper in `scripts/compose.py` (most cases) and direct `svgutils.compose` for full control.
60
61	### Recipe A: helper (recommended)
62
63	```bash
64	uv run --with svgutils --with lxml python scripts/compose.py panels-config.json -o figure.svg
65	```
66
67	`panels-config.json` schema:
68
69	```json
70	{
71	"width_mm": 183,
72	"height_mm": 120,
73	"journal": "nature",
74	"panels": [
75	{"id": "A", "src": "panels/spectrum.svg", "x_mm": 0, "y_mm": 0, "scale": 0.5, "label": "A"},
76	{"id": "B", "src": "panels/topomap.svg", "x_mm": 92, "y_mm": 0, "scale": 0.5, "label": "B"},
77	{"id": "C", "src": "panels/timecourse.svg", "x_mm": 0, "y_mm": 60, "scale": 1.0, "label": "C", "width_mm": 183}
78	]
79	}
80	```
81
82	Panel labels (A, B, C) are placed at top-left of each panel in 12 pt bold sans-serif.
83
84	### Recipe B: direct svgutils (full control)
85
86	```python
87	from svgutils.compose import Figure, SVG, Panel, Text
88
89	fig = Figure(
90	"183mm", "120mm",
91	Panel(
92	SVG("

People who liked this also viewed

6 picks

Type

scientific-figure

37stars7forks

View on GitHub

How to install scientific-figure?

neuromechanist/research-skills/scientific-figure

$npx -y skills add neuromechanist/research-skills --skill scientific-figure

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Scientific Figure
2
3	Compose publication-quality multi-panel figures at exact journal dimensions (mm/pt), validate font sizes against journal minima before export, and write PDF/PNG that the journal will accept without resize.
4
5	## Why this skill exists
6
7	The previous react-pdf workflow had two structural failures:
8
9	1. Composition did not respect physical size. Flexbox layout shifted dimensions in subtle ways during render.
10	2. Fonts shrank below readable thresholds. When content overflowed, react-pdf uniformly scaled the entire figure, sometimes pushing axis labels under 4 pt.
11
12	This skill replaces that workflow. Panels are placed at exact mm coordinates, text is preserved as SVG `<text>` elements so font sizes are inspectable before export, and the validator rescales individual panels rather than the whole figure when a font minimum is violated.
13
14	## Pipeline
15
16	```
17	1. Plan 2. Build elements 3. Compose 4. Validate fonts 5. Export
18	(journal (matplotlib/seaborn (svgutils, at (per-element pt vs (Inkscape if
19	size, panel per panel; SVG out; exact mm/pt; journal minimum; present;
20	grid) optional icons) text preserved) rescale panel if cairosvg
21	below minimum) otherwise)
22	```
23
24	Every step uses on-the-fly execution via `uv run --with` so no permanent installs are required for the skill itself.
25
26	## Step 1: Plan the figure
27
28	Before generating anything, fix the following:
29
30	- Target journal — sets the canvas width. See `references/journal-specs.md` for the full table; the four most common are summarized below.
31	- Panel grid — 1x1, 1x2, 2x2, wide-top + sub-panels, or freeform mm coordinates.
32	- Color palette — pick from `references/color-palettes.md` and reuse across all panels.
33
34	### Journal width / font minimum cheat sheet
35
36	\| Journal \| 1 column \| 2 column \| Min body font \| Notes \|
37	\|---\|---\|---\|---\|---\|
38	\| Nature \| 89 mm \| 183 mm \| 5 pt \| 8 pt for panel labels \|
39	\| Science \| 55 mm \| 120 mm \| 6 pt \| Myriad/Helvetica preferred \|
40	\| Cell \| 85 or 112 mm \| 174 mm \| 6 pt \| Max 225 mm tall \|
41	\| PNAS \| 87 mm \| 180 mm \| 6 pt \| No label below 2 mm tall \|
42
43	The validator in step 4 enforces these. Pick the journal early so the validator can warn early.
44
45	## Step 2: Build the elements
46
47	Each panel is generated independently as an SVG. Common sources:
48
49	- Plots: matplotlib/seaborn, saved with `bbox_inches='tight', transparent=True, format='svg'`. See the `[[plot-styling]]` skill for the library decision tree and SciencePlots recipes.
50	- Icons: transparent PNGs from the `[[transparent-icons]]` skill.
51	- Schematics: for new Python-driven work use `[[svg-primitives]]` (auto-fit boxes, validated arrows, layered z-order); for hand-authored SVG patterns see `[[svg-figure]]`.
52
53	Save each element to a working directory (typically `panels/`), then compose them in step 3.
54
55	## Step 3: Compose the figure
56
57	The composer is built around [svgutils](https://svgutils.readthedocs.io/) (MIT license; `uv run --with svgutils`). It places panels at exact mm coordinates and preserves text as inspectable SVG `<text>` elements.
58
59	Two ways to compose: the `Figure` helper in `scripts/compose.py` (most cases) and direct `svgutils.compose` for full control.
60
61	### Recipe A: helper (recommended)
62
63	```bash
64	uv run --with svgutils --with lxml python scripts/compose.py panels-config.json -o figure.svg
65	```
66
67	`panels-config.json` schema:
68
69	```json
70	{
71	"width_mm": 183,
72	"height_mm": 120,
73	"journal": "nature",
74	"panels": [
75	{"id": "A", "src": "panels/spectrum.svg", "x_mm": 0, "y_mm": 0, "scale": 0.5, "label": "A"},
76	{"id": "B", "src": "panels/topomap.svg", "x_mm": 92, "y_mm": 0, "scale": 0.5, "label": "B"},
77	{"id": "C", "src": "panels/timecourse.svg", "x_mm": 0, "y_mm": 60, "scale": 1.0, "label": "C", "width_mm": 183}
78	]
79	}
80	```
81
82	Panel labels (A, B, C) are placed at top-left of each panel in 12 pt bold sans-serif.
83
84	### Recipe B: direct svgutils (full control)
85
86	```python
87	from svgutils.compose import Figure, SVG, Panel, Text
88
89	fig = Figure(
90	"183mm", "120mm",
91	Panel(
92	SVG("

People who liked this also viewed

6 picks

Type