vega_skill

Overview

This skill generates data-driven charts using Vega-Lite for common visualizations and Vega for advanced, custom visuals. It converts arrays of objects into publish-ready JSON specs you can paste inside fenced vega-lite or vega blocks. Use Vega-Lite for most needs and Vega when you require radar charts, word clouds, or force-directed layouts.

How this skill works

You provide data as an array of objects and map fields to encodings like x, y, color, and size. The skill validates and emits a valid JSON spec that always includes the appropriate $schema and correct data types (quantitative, nominal, ordinal, temporal). It favors Vega-Lite by default and switches to Vega when the chosen visualization requires features beyond Vega-Lite.

When to use it

• Plotting bar, line, scatter, area, heatmap, or multi-series analytics from numeric arrays.
• Creating statistically accurate visualizations where field typing and encodings matter.
• Building interactive, declarative charts with consistent schema and cross-browser compatibility.
• Generating radar charts, word clouds, or force-directed graphs (use Vega).
• Avoid for process diagrams or KPI cards—use mermaid or infographic tools instead.

Best practices

• Always include the correct $schema for Vega-Lite or Vega at the top of the JSON spec.
• Supply data as an array of objects and ensure field names match exactly (case-sensitive).
• Use valid JSON with double quotes and no trailing commas; validate before rendering.
• Choose Vega-Lite for 90% of charts; reserve Vega for specialized layouts like radar or word clouds.
• Set field types explicitly to quantitative, nominal, ordinal, or temporal to avoid inference errors.

Example use cases

• Multi-series sales dashboard: multi-line chart with independent y-scales for different metrics.
• Exploratory data analysis: scatter plots with size and color encoding to surface clusters.
• Heatmap of correlation matrices for quick pattern recognition across features.
• Radar chart of product attributes or a word cloud of user feedback (Vega).
• Publication-quality charts where schema compliance and reproducible specs are required.