4 tagged tableau, measured the same way as everything else here.
Agent Claude Code
Turns an insights narrative into a concrete dashboard design spec — metric selection, layout grid, chart types, color plan, and linter evaluation criteria. Stage 2 of the dashboard pipeline. Produces no XML.
Agent Claude Code
Scores a rendered dashboard screenshot against the design-standards rubric and the run's design spec, producing a lint report with XML-level fixes. Stage 4 of the dashboard pipeline — run after the human drops a render screenshot.
Agent Claude Code
Analyzes a dataset against a business question using Python, producing an analytics plan and a standalone insights narrative. Use as stage 1 of the dashboard pipeline, or on its own when the user only wants analysis.