Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› aaag-tables-figures
aaag-tables-figures
GitHub用于为《美国地理学家协会公报》稿件设计地图、图表和表格。提供制图标准（投影、分类、无障碍色彩）、遥感与统计图规范，以及展品决策审查，确保符合期刊严格的制图质量和字数限制。
Trigger Scenarios
设计或修改地图、遥感图、模型图或表格
审稿人/编辑指出地图投影、分类、图例或可读性问题
精简展品以符合包含图注在内的11,000字上限
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill aaag-tables-figures -g -y
SKILL.md
Frontmatter
{
"name": "aaag-tables-figures",
"description": "Use when building maps, figures, and tables for an Annals of the American Association of Geographers manuscript. The Annals has a dedicated Cartography Editor and Graphics Guidelines; maps are held to cartographic standards, not just plotted. Improves exhibits; it does not run the analysis."
}
Maps, Figures & Tables (aaag-tables-figures)
The Annals is a geography journal: maps are arguments, and the journal holds them to cartographic standards. There is a Cartography Editor and Graphics Guidelines (检索于 2026-06；以官网为准), and every exhibit counts toward the 11,000-word cap (figure captions are included). Make each exhibit carry its weight and read correctly in print and grayscale.
When to trigger
- Designing or revising maps, remote-sensing figures, model plots, or tables
- A reviewer/editor flagged a map's projection, classification, legend, or legibility
- Trimming exhibits to fit the inclusive word cap
Cartographic essentials (maps are held to standard here)
- Projection chosen on purpose — appropriate to area/distance/shape for the analysis; name it.
- Classification stated — method (quantiles, equal-interval, natural breaks) and class count, with a rationale; the choropleth message must not be an artifact of the break scheme.
- Map elements present — scale bar, north arrow where needed, legend, data source, and date; graticule/inset where it aids reading.
- Normalize areal data — rates/densities, not raw counts, on choropleths; avoid MAUP-driven artifacts.
- Accessible color — colorblind-safe, sequential/diverging schemes matched to the data type; legible in grayscale; sufficient figure-ground contrast.
Remote-sensing & physical figures
- State band combination/stretch; include a spatial error/uncertainty panel where accuracy matters.
- Give coordinates/extent and resolution so the figure is reproducible and locatable.
Model & statistical figures
- Prefer effect/marginal-effect, GWR-surface, and event/trend plots over dense tables; show uncertainty bands. Tables carry effect sizes and CIs, not stars alone.
- Keep model figures spatially interpretable: label the spatial unit, neighborhood definition, bandwidth, buffer, or grid resolution that drives the pattern. If the figure is a map of residuals, clusters, or local coefficients, add the diagnostics that prevent readers from treating noise as geography.
Exhibit decision gate
Use this gate before adding or keeping any exhibit. The Annals page budget is tight, and geography reviewers will ask whether the visual does analytic work rather than decorating the manuscript.
| Exhibit type | Keep only if it answers | Required audit trail |
|---|---|---|
| Locator or study-area map | Does the reader need spatial context to understand the claim? | Projection, extent, scale, source, date, and why the inset or boundary was chosen |
| Choropleth / thematic map | Is the mapped pattern central to the argument? | Denominator, classification method, class count, sensitivity to an alternate break scheme |
| Remote-sensing / imagery panel | Does the image identify land cover, change, or measurement uncertainty? | Sensor/product, bands or index, acquisition date/window, resolution, accuracy or validation note |
| Model plot | Does the plot reveal effect size, heterogeneity, or uncertainty better than a table? | Estimand, sample, confidence interval, spatial/temporal unit, robustness pointer |
| Table | Is exact comparison necessary and too dense for prose? | Slim column set, effect sizes/CIs, notes defining variables and sample restrictions |
Decision rule: if an exhibit cannot state its analytic claim in one sentence, cut it or move it to supplementary material. If it repeats a result already clear in text, replace it with a smaller table or one caption sentence.
Cartography QA handoff
Before final submission, create a one-page exhibit QA note for coauthors or the Cartography Editor pathway. It should list each exhibit, its file name, source data, projection, classification, color scheme, print-size legibility check, and word-budget cost. Mark any item as HOLD if the map is not reproducible from archived data or if the caption omits what/where/when/source/takeaway.
For multi-panel figures, do a panel-level audit rather than approving the composite as a whole: each panel needs a label, unit, legend logic, and a reason to remain. Avoid mixing raw maps, model effects, and uncertainty diagnostics in one panel stack unless the caption explains the reading order.
Self-containment & house style
- Every exhibit is readable standalone: caption states what, where, when, source, and the takeaway.
- Follow the Graphics Guidelines for resolution, fonts, and file format; vector for line art, high-resolution raster for imagery (verify current specs on the page).
- Count captions, tables, and figures toward the 11,000-word cap — budget exhibits deliberately.
Common map failure → fix (what the Cartography Editor flags)
| Failure | Why it misleads | Fix |
|---|---|---|
| Raw counts on a choropleth | bigger/more-populous units always look "high" | map rates/densities; normalize |
| Default Web Mercator for area analysis | distorts area at high latitudes | choose an equal-area projection; name it |
| Auto "natural breaks" with no rationale | the message is an artifact of the break scheme | justify method + class count; test alternatives |
| Rainbow palette | not colorblind-safe; false ordering | sequential/diverging colorblind-safe scheme |
| No scale bar / source / date | unverifiable, not reproducible | add required map elements |
| Tiny multiples illegible in print | reviewer cannot read the result | resize, simplify, or split exhibits |
Checklist
- Projection named and appropriate to the analysis
- Classification method + class count stated and justified
- Scale bar / legend / source / date present; areal data normalized
- Colorblind-safe and grayscale-legible; adequate contrast
- RS figures: band/stretch + accuracy/error panel; extent + resolution given
- Each exhibit passes the keep/cut gate and states its analytic claim in one sentence
- Model maps/plots name the estimand, spatial unit, uncertainty, and robustness pointer
- Exhibit QA note records file name, data source, projection, classification, palette, and print check
- Captions self-contained; exhibit budget fits the inclusive word cap
- Graphics meet the journal's resolution/format specs (verify)
Anti-patterns
- Default-software projection and auto-classification with no rationale (artifactual choropleth)
- Raw counts on a choropleth instead of rates/densities
- Rainbow / non-colorblind-safe palettes; maps illegible in grayscale
- Missing scale bar, legend, or data source/date
- Overrunning the word cap because captions and tables were not counted
Output format
【Exhibit】map / RS figure / model plot / table
【Cartography】projection + classification + required elements? [Y/N]
【Accessibility】colorblind-safe + grayscale-legible? [Y/N]
【Self-contained】caption states what/where/when/source/takeaway? [Y/N]
【Analytic claim】one-sentence visual claim + keep/cut decision
【QA handoff】file/data/projection/classification/palette/print check complete? [Y/N]
【Word-cap budget】exhibits + captions counted? [Y/N]
【Next】aaag-writing-style
Supplementary resources
../../resources/external_tools.md— GIS/cartography and figure tooling (QGIS, ggplot2/tmap, matplotlib)../../resources/official-source-map.md— Graphics Guidelines and Cartography Editor
Version History
- 1839142 Current 2026-07-05 12:23
