Changelog

0.2.2

• Added a portable plugin manifest, listing text, and EC icons.
• Clarified user overrides, tool requirements, and skipped checks. Review can use an allowed separate reviewer or self-review.
• Added a reproducible release ZIP builder and package tests.
• Updated installation guidance and added an acceptance checklist.

0.2.1

• Removed check_data.py. In test runs it raised false alarms and added nothing agents didn't already catch; the data checks are back to short written rules.
• check_chart.py now flags text on a line where a gray context line meets its accent-colored part, and text crossing a zero or reference line.

0.2.0

• check_data.py checks a data table before charting: total rows mixed with members, duplicate keys, trailing zero periods, status flags, sentinel codes, mixed units, gaps, frozen values.
• check_svg.py runs the overlap, clipping, text size, contrast, and color-blindness checks on SVG exports from Plotly, Vega-Lite/Altair, ggplot2, and D3 (needs Chrome).
• check_chart.py adds number-format (offset text like "1e6", fractional year ticks, mixed decimals), category-order (unsorted bars, lines across categories), segment-edges (touching fills with no border), and invisible-arrow (annotation arrows the house style hides).
• New rules: margins of error for poll leads, both absolute levels for relative risks, fitted lines inside the data, gaps in lines for missing periods, white edges between filled segments.

0.1.0

First release.

• Workflow for explanatory charts: check the data, write a one-line brief, pick the form, build, check, review, deliver.
• Rules tagged by evidence ([E], [P], [T]), with citations for every rule ID in references/sources.md.
• check_chart.py lints matplotlib charts: overlapping or clipped text, labels on data, bar baselines, dual axes, color-blind confusable series, contrast, plot shape, and more.
• check_palette.py tests palettes for color-blind safety and contrast.
• evident.py helper with presets for blog, X, LinkedIn, slides, reports, and phones.
• A visual review loop on the rendered chart before delivery, up to three rounds.
• Chart critique with ranked, rule-cited fixes.
• House themes for matplotlib, ggplot2, Plotly, Vega-Lite, and D3.
• Installs for Claude Code, Codex, Gemini CLI, and any agent that npx skills supports.