---
name: data-visualization
description: Choose, build and validate charts or dashboards from data. Use when the task involves analytical encoding, uncertainty, monitoring, comparisons or narrative claims about data.
---

# Data visualization

Write the analytical question and data source before choosing marks. Check units, missingness, aggregation and uncertainty. Prefer Observable Plot or Vega-Lite for standard charts; use D3/SVG for bespoke geometry. Keep essential values visible without hover and preserve a mobile reading order. Test chart semantics, source disclosure, keyboard access and export. `fixtures/chart` demonstrates a declarative chart; use `schemas/visual-contract.schema.json` and the shared QA protocol.