datavizlisted
Install: claude install-skill hsb3/dotfiles-agents
# Data visualization
Design charts, graphs, plots, and dashboards that are **correct, legible, and honest** — in any
medium and any library. This skill is consulted **before** any visualization is coded: the choice
of mark, scale, color, and layout is decided here, then the drawing library just executes it.
**Read this hub, then pull the reference that matches the decision you're on.** The rules are
library-agnostic; `references/libraries.md` has the per-tool reminders.
## Scope and boundaries
One owner per fact — these boundaries keep skills from fighting over triggers:
- **This skill owns anything that encodes a dataset** — bar/line/scatter/area, distributions,
part-of-whole, relationship, geospatial, dashboards, KPI/stat tiles, sparklines, heatmaps,
small multiples. If the visual carries numbers, it is this skill's job.
- **Structural diagrams** (boxes, arrows, flow, containment, architecture, sequence, ER, org,
process) → the `diagrams` and `mermaid` skills. Do not use a chart library for structure, and
do not use diagram tools (or Mermaid `pie`/`xychart`) for real data.
- **Deck/slide theming and PowerPoint charts** → `presentations`. A chart on a slide inherits the
deck's theme tokens; don't bake a chart palette into a slide.
## Workflow — before you draw
1. **State the takeaway in one sentence.** That sentence is the chart title. If you can't write
it, the chart isn't ready.
2. **Name the data shape** — number of series, number of categories, whethe