Data Visualization

This is the opinionated data viz guide you need when Claude is generating charts. It walks through chart selection (what to use when, why pie charts almost always suck), color theory for sequential and diverging palettes, axes rules like starting bar charts at zero, and annotation best practices. Includes executable examples via the belt CLI for line charts, horizontal bars, and KPI cards using matplotlib and HTML rendering. The chart selection table alone is worth having around: it maps data relationships to chart types and explicitly calls out what not to use. If you're building dashboards, reports, or any data presentation and want Claude to follow actual design principles instead of making rainbow pie charts, load this.

Install to Claude Code

npx -y skills add inference-sh/skills --skill data-visualization --agent claude-code