figures

Name: figures
Author: scdenney

34stars0forks

How to install figures?

scdenney/open-science-skills/figures

$npx -y skills add scdenney/open-science-skills --skill figures

Installs into the current project.

Summary

Design and format publication-quality figures. Use for chart choice, color, scales, legends, captions, accessibility, and reproducible figure workflows.

Files · 1

View on GitHub

SKILL.md

1	# Figure Designer
2
3	## Heritage and scope
4
5	This is an original Open Science Skills workflow for figure production in social-science manuscripts. It is general — apply to any figure type (line, bar, point, density, map, network, small multiples). For figures whose interpretation depends on method-specific standards, also consult the relevant sibling skill (`conjoint-design`, `conjoint-diagnostics`, `list-experiment`, `topic-modeling`, `text-classification`, `vlm-ocr-pipeline`). For end-stage QA on a finished figure set, hand off to `figure-table-audit`.
6
7	A good figure earns its place in the manuscript: it makes a single comparison legible, it can be read without the surrounding text, and it can be regenerated from a script. If a figure cannot do all three, it is not yet ready.
8
9	## Instructions
10
11	### 1. State the comparison before drawing
12
13	Write down, in one sentence, what the figure is supposed to let the reader see. Examples:
14
15	- "Treatment effect is larger among low-information respondents than among high-information respondents."
16	- "Vote share for Party A declines monotonically with district urbanization."
17	- "Topic 7 prevalence rises sharply after the 2024 election."
18
19	If you cannot state the comparison in one sentence, the figure has too many goals — split it into multiple panels or multiple figures.
20
21	### 2. Choose the chart type from the comparison
22
23	Match the geometry to the comparison, not to the data type:
24
25	- Comparing magnitudes across a small number of categories: bar chart (or dot/lollipop for many categories).
26	- Tracking a quantity over a continuous variable: line chart for ordered/continuous x; point + ribbon for uncertainty.
27	- Showing distribution: histogram (single), density (compare a few), boxplot/violin (compare many).
28	- Showing relationship between two continuous variables: scatter, with trend line + uncertainty band if a model is implied.
29	- Showing effect estimates: coefficient plot (point + CI), preferred over tables for visual comparison across models.
30	- Showing many small comparisons: small multiples (faceting) on a shared scale, not multiple separate figures.
31
32	Avoid pie charts, 3D anything, dual-axis, and donut charts in academic figures.
33
34	### 3. Pick scales and axes deliberately
35
36	- Always label both axes with the quantity AND the units. "Income" is wrong; "Household income (USD, 2020)" is right.
37	- For percentages, write `0.00–1.00` or `0%–100%` consistently — do not mix within a figure.
38	- Use a log scale when the data span more than two orders of magnitude OR when a multiplicative effect is the substantive story; label the scale as log.
39	- Anchor axes at zero ONLY when the zero is meaningful for the comparison (counts, proportions). For deviations, differences, and z-scores, do not force a zero baseline.
40	- Use the same y-axis range across panels of small multiples unless the claim is explicitly about within-panel scale.
41	- Reverse the y-axis only when the substantive direction demands it (e.g., rank where 1 is "best") and label the reversal in the caption.
42
43	### 4. Choose color with intent
44
45	- Sequential (light → dark): for ordered/quantitative variables (income brackets, time, density).
46	- Diverging (red ↔ blue, etc.): for variables with a meaningful midpoint (effect sign, deviation from baseline).
47	- Categorical: for unordered groups; use a colorblind-safe palette (Okabe-Ito, viridis discrete, ColorBrewer Set2/Dark2).
48	- Test grayscale: convert the figure and check that the comparison still reads. If it does not, add shape, line type, or direct labels.
49	- Do not encode the same variable in both color and shape unless reinforcing for accessibility — that is fine; just keep the legend honest.
50	- Default `rainbow` / `jet` ramps are a publication smell; replace them.
51
52	### 5. Match legend order to the visual order in the chart
53
54	Legends are read alongside the plot, so the legend order must mirror the data's visual order — readers should never have to scan back and forth to decode a series:
55
56	- For lines, areas, or stacked elements arranged top-to-bottom in the plot at the rightmost x-value, list legend entries in the same top-to-bottom order.
57	- For bars or categories arranged left-to-right, use a horizontal legend listed left-to-right in the same order.
58	- For panels (small multiples), follow the panel grid: rows top-to-bottom, columns left-to-right.
59	- Do NOT alphabetize the legend when the data have a natural order (time, magnitude, treatment intensity, ordinal scale). Alphabetical legends are a frequent cause of misreading.
60	- Prefer direct labeling at the line end / bar end when there are five or fewer series — a labelled line beats any legend.
61	- When a legend is unavoidable, place it where it does not occlude data, and match colors and line styles between figure and legend

People who liked this also viewed

6 picks

Type

find-skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skSkillsJul 20262.6M27k
grill-meA relentless interview to sharpen a plan or design.SkillsJul 2026633k182k
vercel-react-best-practicesVercel's official collection of agent skillsSkillsJul 2026573k29k
improve-codebase-architectureScan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.SkillsJul 2026522k182k
setup-matt-pocock-skillsConfigure this repo for the engineering skills — set up its issue tracker, triage label vocabulary, and domain doc layout. Run once before first use of the other engineering skills.SkillsJul 2026454k182k
triageMove issues and external PRs through a state machine of triage roles — categorise, verify, grill if needed, and write agent-ready briefs.SkillsJul 2026424k182k

figures

34stars0forks

View on GitHub

How to install figures?

scdenney/open-science-skills/figures

$npx -y skills add scdenney/open-science-skills --skill figures

Installs into the current project.

Summary

Design and format publication-quality figures. Use for chart choice, color, scales, legends, captions, accessibility, and reproducible figure workflows.

Files · 1

View on GitHub

SKILL.md

1	# Figure Designer
2
3	## Heritage and scope
4
5	This is an original Open Science Skills workflow for figure production in social-science manuscripts. It is general — apply to any figure type (line, bar, point, density, map, network, small multiples). For figures whose interpretation depends on method-specific standards, also consult the relevant sibling skill (`conjoint-design`, `conjoint-diagnostics`, `list-experiment`, `topic-modeling`, `text-classification`, `vlm-ocr-pipeline`). For end-stage QA on a finished figure set, hand off to `figure-table-audit`.
6
7	A good figure earns its place in the manuscript: it makes a single comparison legible, it can be read without the surrounding text, and it can be regenerated from a script. If a figure cannot do all three, it is not yet ready.
8
9	## Instructions
10
11	### 1. State the comparison before drawing
12
13	Write down, in one sentence, what the figure is supposed to let the reader see. Examples:
14
15	- "Treatment effect is larger among low-information respondents than among high-information respondents."
16	- "Vote share for Party A declines monotonically with district urbanization."
17	- "Topic 7 prevalence rises sharply after the 2024 election."
18
19	If you cannot state the comparison in one sentence, the figure has too many goals — split it into multiple panels or multiple figures.
20
21	### 2. Choose the chart type from the comparison
22
23	Match the geometry to the comparison, not to the data type:
24
25	- Comparing magnitudes across a small number of categories: bar chart (or dot/lollipop for many categories).
26	- Tracking a quantity over a continuous variable: line chart for ordered/continuous x; point + ribbon for uncertainty.
27	- Showing distribution: histogram (single), density (compare a few), boxplot/violin (compare many).
28	- Showing relationship between two continuous variables: scatter, with trend line + uncertainty band if a model is implied.
29	- Showing effect estimates: coefficient plot (point + CI), preferred over tables for visual comparison across models.
30	- Showing many small comparisons: small multiples (faceting) on a shared scale, not multiple separate figures.
31
32	Avoid pie charts, 3D anything, dual-axis, and donut charts in academic figures.
33
34	### 3. Pick scales and axes deliberately
35
36	- Always label both axes with the quantity AND the units. "Income" is wrong; "Household income (USD, 2020)" is right.
37	- For percentages, write `0.00–1.00` or `0%–100%` consistently — do not mix within a figure.
38	- Use a log scale when the data span more than two orders of magnitude OR when a multiplicative effect is the substantive story; label the scale as log.
39	- Anchor axes at zero ONLY when the zero is meaningful for the comparison (counts, proportions). For deviations, differences, and z-scores, do not force a zero baseline.
40	- Use the same y-axis range across panels of small multiples unless the claim is explicitly about within-panel scale.
41	- Reverse the y-axis only when the substantive direction demands it (e.g., rank where 1 is "best") and label the reversal in the caption.
42
43	### 4. Choose color with intent
44
45	- Sequential (light → dark): for ordered/quantitative variables (income brackets, time, density).
46	- Diverging (red ↔ blue, etc.): for variables with a meaningful midpoint (effect sign, deviation from baseline).
47	- Categorical: for unordered groups; use a colorblind-safe palette (Okabe-Ito, viridis discrete, ColorBrewer Set2/Dark2).
48	- Test grayscale: convert the figure and check that the comparison still reads. If it does not, add shape, line type, or direct labels.
49	- Do not encode the same variable in both color and shape unless reinforcing for accessibility — that is fine; just keep the legend honest.
50	- Default `rainbow` / `jet` ramps are a publication smell; replace them.
51
52	### 5. Match legend order to the visual order in the chart
53
54	Legends are read alongside the plot, so the legend order must mirror the data's visual order — readers should never have to scan back and forth to decode a series:
55
56	- For lines, areas, or stacked elements arranged top-to-bottom in the plot at the rightmost x-value, list legend entries in the same top-to-bottom order.
57	- For bars or categories arranged left-to-right, use a horizontal legend listed left-to-right in the same order.
58	- For panels (small multiples), follow the panel grid: rows top-to-bottom, columns left-to-right.
59	- Do NOT alphabetize the legend when the data have a natural order (time, magnitude, treatment intensity, ordinal scale). Alphabetical legends are a frequent cause of misreading.
60	- Prefer direct labeling at the line end / bar end when there are five or fewer series — a labelled line beats any legend.
61	- When a legend is unavoidable, place it where it does not occlude data, and match colors and line styles between figure and legend

People who liked this also viewed

6 picks

Type