Display Quantitative Information

Use this skill to help an agent create, critique, redesign, audit, or explain quantitative displays that let people reason from evidence. The default standard is truthfulness first, comparison power second, and visual economy third. Minimalism is not the goal; clear quantitative reasoning is.

Activation boundaries

Use this skill for chart choice, data visualization code, dashboard review, statistical/scientific figures, misleading graphics, graph redesign, tables used as evidence, uncertainty displays, small multiples, map-based quantitative displays, or user language such as data-ink, chartjunk, graphical integrity, lie factor, Tufte, visual evidence, publication-ready figure, or dashboard critique.

Do not use it for decorative illustration, infographics with no measured quantities, brand-only design, slide aesthetics without data, or general data wrangling unless a display or visual explanation is part of the task.

Working loop

1. Name the viewer's task: lookup, comparison, trend, relationship, distribution, part-to-whole, geography, uncertainty, monitoring, explanation, or persuasion.
2. Inspect the data structure: grain, units, denominators, time order, grouping, spatial structure, missingness, transformations, sample size, and uncertainty.
3. Choose the display from the task and data, not from a favorite chart type. Use references/display-selection.md when the choice is not obvious.
4. Audit integrity before aesthetics: baselines, scales, proportionality, encodings, transformations, omitted context, denominators, uncertainty, source, and accessibility. Use references/integrity-audit.md or scripts/audit_visual_display.py for structured specs.
5. Redesign by improving the intended comparison. Remove distracting marks, but keep labels, notes, reference lines, captions, and structure when they help interpretation.
6. Deliver the artifact requested: chart, code, SVG, design spec, critique, dashboard review, or short recommendation. Put the highest-impact fix first.