Catalog backlog: coverage burn-down, light/dark stragglers, sub-70 regens, repo hygiene

[MarkusNeusinger]

Catalog findings from the 2026-07-08 audit (recounted 2026-07-09):

• Medium#25 — 148 specs still Python-only; 261/324 (80.6%) below full 15-library coverage (distribution: 8 impls ×135, 9×50, 10×40, 15×63). Resume targeted (not all) regen via babysit-pipeline; burn down the 8/9/10-impl buckets. Also still pending from the stopped June queue: stereonet-equal-area, column-stratigraphic, heatmap-cohort-retention.
• Medium#24 — flagship 3D/geo specs are the sparsest (scatter-3d, surface-basic: 7 Python-only impls each, no makie — the library that would shine there). Target the 3D cluster first (makie, echarts, ggplot2); record legitimate per-library exclusions explicitly.
• Medium#20 — last 3 old-design metadata files without light/dark previews: line-stress-strain (bokeh, letsplot), heatmap-cohort-retention (plotnine). Three targeted impl-generate runs finish the migration at 100%.
• Low#22 — ten sub-70 approved impls (7× pygal faking contours/maps/trees, 2× altair, 1× seaborn — exact list in the audit). Regenerate with opus per the migration playbook, or introduce explicit per-library opt-outs (relates to AR-06 handling in Quality-score calibration: first-review score, AR sentinel, worst-first regen, anchors #9626).
• Low#16 — git-tracked strays: plots/bar-basic/implementations/python/plot-{dark,light}.png (~200 KB each), plots/contour-3d/implementations/python/_run_plotly.py, 10 stale .gitkeep files (5 specs × 2 dirs). git rm --cached; grep automation/ before deleting the runner.
• Low#18 — 118 specs with updated: null + inconsistent date quoting; one-shot codemod (updated := created).
• Low#14 — search_no_results shows real content gaps: "backtest equity curve" spec is definitively absent; arrow/flow-map is likely a tagging/discoverability issue (flowmap-origin-destination exists). File the spec request; fix tagging.
• Low#19 — duplicate-adjacent spec clusters (regression-linear/-polynomial/-lowess; sma/ema) — cross-check per-spec traffic before merging siblings.

Audit: agentic/audits/2026-07-08-product-ux.md.