Plot Data

Render with deterministic plotting rules.

• Use matplotlib for chart generation.
• Always use uv for Python-related operations.
• Prefer Python standard library plus matplotlib; avoid adding dependencies unless strictly required.
• Use readable defaults:
  • figure size appropriate for the data density
  • axis labels and title when available
  • legible tick density and rotation for long labels
  • visible grid for quantitative charts unless user asks otherwise

Produce deterministic artifacts.

• Save outputs to output/plot-data.
• If directory does not exist, create it using Python standard library (os).
• Generate a unique basename:
  • plot_<chart-type>_<topic-slug>_<uuid8>
• Save:
  • <basename>.png as the final chart image
  • <basename>.json metadata with rendering context
• Metadata JSON must include:
  • request: inferred user intent and options
  • chart: chart type, title, axis labels, styling choices
  • data_summary: row count, column names, dropped/cleaned rows count
  • artifacts: absolute or workspace-relative artifact paths
  • generated_at: ISO-8601 timestamp in UTC

Return concise user-facing output.

• Provide a short summary of what was plotted.
• Include the artifact path(s) for PNG and metadata JSON.
• If assumptions were made (for example chart-type default), state them briefly.

Handle failures explicitly.

• If plotting fails due to invalid data shape, report the specific mismatch.
• If execution fails due to transient runtime issues, retry once.
• Never fabricate data points to make the plot render.

Clean up temporary files only.

• Remove temporary scripts and intermediate files created during plotting.
• Do not delete final artifacts inside output/plot-data.