fix(stats): count coverage against all 15 libraries

[MarkusNeusinger]

Summary

• Fixes the user-reported "Polar Scatter Plot: 15/9" on https://anyplot.ai/stats — the coverage matrix divided by a hardcoded 9 (a leftover from when nine libraries were supported), so a fully covered spec rendered as 15/9 and the "of N possible" total used the wrong denominator. The API already computes coverage_percent against len(SUPPORTED_LIBRARIES) (15), so only the page was wrong.
• /insights/dashboard now serves total_libraries and StatsPage renders the matrix against it (falling back to library_stats.length for a frontend-before-API deploy), so the denominator tracks the library set instead of drifting from it.
• Reworks the cells for legibility, as requested in the same report: a 16 px mark (was 10 px) centred in a 24 px link so the target meets WCAG 2.2 SC 2.5.8, full coverage as solid brand green and anything below it carrying an ink outline (dashed when a spec has no implementation at all), a labelled three-state legend instead of the less/more opacity ramp, and · N below 15/15 in the summary line. The interesting signal is which specs are not complete, and those are the minority the old opacity ramp buried.

Test plan

• uv run pytest tests/unit/api -q — 794 passed; the dashboard test now asserts total_libraries == len(SUPPORTED_LIBRARIES).
• yarn vitest run src/pages/StatsPage.test.tsx src/pages/DebugPage.test.tsx — green, with three new cases: the summary line renders 987 of 2130 possible · 1 below 15/15, cells are labelled Basic Scatter Plot: 1/15 / Basic Line Plot: 15/15, and a payload without total_libraries still derives 15 from library_stats.
• Gates: npx tsc --noEmit, eslint, prettier, uv run ruff check, uv run ruff format --check, uv run mypy api/routers/insights.py — all clean.
• Visual: drove /stats in headless Chromium against a mocked dashboard (121 complete specs + 3 partial + 1 empty) at 1280×900 and 390×844, light and dark. Summary read 97.4% · 1836 of 1875 possible · 4 below 15/15; the four outliers are immediately pickable out of the strip in all four combinations. A live run against the production API was not possible here (no DB credentials in the session).

Checklist

• changelog.d/stats-coverage-denominator.md added (Fixed + Changed, no CHANGELOG.md bullet)
• Related documentation checked — docs/reference/api.md describes /insights/dashboard without a field list, so the new response field needs no doc change.

🤖 Generated with Claude Code

https://claude.ai/code/session_01AkKTuCSmiEeUaPQDK5kXqS

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[Copilot]

🟡 Changes recommended

The redesigned cells have insufficient touch-target sizing and light-theme contrast, and the compatibility fallback lacks regression coverage.

Get a fresh assessment by requesting another Copilot review.

Pull request overview

Fixes stats-page coverage calculations to use all supported libraries and improves incomplete-coverage visibility.

Changes:

• Adds total_libraries to the dashboard API.
• Updates coverage totals, cells, legend, and tests.
• Documents the fix and visual changes.

File summaries

File	Description
api/routers/insights.py	Exposes the canonical library count.
app/src/pages/StatsPage.tsx	Corrects coverage rendering and redesigns cells.
app/src/pages/StatsPage.test.tsx	Tests 15-library totals and labels.
tests/unit/api/test_routers.py	Verifies the API field.
changelog.d/stats-coverage-denominator.md	Records the changes.

Review details

• Files reviewed: 5/5 changed files
• Comments generated: 3
• Review effort level: Balanced

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.