What are the best data visualization skills for Claude Code?

Ask an agent for a chart without guidance and you get default matplotlib: wrong chart type, rainbow palette, unreadable at publication size. The fix is a visualization skill, and the catalog's leader is data-visualization from anthropics at 9,375 installs (skills.sh registry).

Key takeaways

• data-visualization by anthropics is the category leader at 9,375 installs, from the knowledge-work-plugins repo with 22,524 GitHub stars (skills.sh registry, GitHub).
• Chart-type selection is the highest-leverage thing these skills encode; wrong chart type is the most common failure in agent-generated figures.
• Web charting has a first-party champion: chart-visualization from antvis, the AntV team, at 4,884 installs.
• Specialists exist per output: documents (chart-designer, 3,342 installs), infographics (1,736 installs), maps (mapbox, 1,252 installs).
• Visualization sits between data and design; the ranked lists at best skills for backend and data and the UI design hub cover both sides.

Why does Claude Code need a data visualization skill?

Because correct-looking charts are easy and correct charts are not. An agent will happily draw a pie chart with twelve slices or plot a trend as grouped bars; nothing in its default behavior enforces the mapping between data relationship and chart form. A visualization skill loads that discipline into context: chart-type selection rules, color theory with accessibility contrast, labeling and scale conventions. The difference is visible in one generation, which is why the category's top skill carries 9,375 installs (skills.sh registry, July 2026).

What is the best data visualization skill overall?

data-visualization from anthropics/knowledge-work-plugins is the clear pick at 9,375 installs, backed by 22,524 GitHub stars on the repo (skills.sh registry, GitHub). It is built around a chart-selection guide - what you are showing mapped to the chart that shows it - plus Python patterns for matplotlib, seaborn, and plotly, with color theory and accessibility baked in.

npx skills add anthropics/knowledge-work-plugins

Strong alternatives: chart-visualization by antvis (4,884 installs) for interactive web charts, and chart-designer (3,342 installs) when the chart's destination is a document or deck rather than an app. The ranked field lives at best skills for backend and data.

Which skill is best for interactive web charts?

chart-visualization from antvis, at 4,884 installs (skills.sh registry). It comes from the team that maintains the AntV charting libraries, so the guidance reflects how production web charts are actually assembled: interaction states, responsive sizing, and data-density decisions that static-figure skills never mention. For chart components inside a React dashboard or a data product, it is the better starting point than the Python-first leader; pair it with the UI skills in the frontend and design category.

What about infographics, documents, and maps?

Each output format has a specialist. infographic-creator (1,736 installs) composes narrative graphics rather than lone charts. chart-designer (3,342 installs, from a 286-star office-skills repo) styles figures for slides and reports. mapbox-data-visualization-patterns (1,252 installs) is mapbox's first-party skill for geospatial visualization, where general chart advice simply does not apply. And data-visualization from owl-listener's designer-skills (994 installs, 1,809 GitHub stars) approaches the same problem from a designer's vocabulary (all counts: skills.sh registry, July 2026).

How do visualization skills relate to dashboard skills?

Visualization skills govern the single figure; dashboard skills govern the page of them. Chart-type selection and palette discipline come from the former, metric hierarchy and layout from the latter, and a serious data product needs both. If your endpoint is a full monitoring or KPI surface, combine one skill from this guide with the picks in best skills for frontend design.

A complete starter stack

• data-visualization (9,375 installs): the theory layer - chart selection, color, accessibility.
• chart-visualization (4,884 installs): the web renderer layer from the AntV team.
• chart-designer (3,342 installs): add when output goes into decks and documents.
• mapbox-data-visualization-patterns (1,252 installs): add for geospatial work.

Common pitfalls

• Letting the agent default to pie charts and rainbow palettes: install the theory skill before generating anything you plan to publish.
• Using Python-figure guidance for web charts: interaction and responsiveness change the rules; use the antvis skill for the web.
• Ignoring accessibility: contrast and colorblind-safe palettes are encoded in the leading skill; do not strip them out for aesthetics.

One theory skill, one renderer skill, and the default-matplotlib era of your repo is over.

Common questions