03 — Adapter Freshness (v_rp_adapter_freshness_status, v_rp_adapter_coverage_dashboard_v2)

Per-source: adapter_exists, last_scan_at, last_success_at, last_error, rows_observed, freshness_age_hours, coverage_status (COVERED / PARTIAL / STALE / STALE_PARTIAL / MISSING_ADAPTER).

Live freshness

• 5 host/FS adapters refreshed 2026-06-05 02:10 UTC (≈6.5h ago) by the systemd orchestrator — all COVERED, status OK: host_crontab 02:10:02 (54), systemd_timers 02:10:08 (22), fs_dot_bin 02:10:12 (287), fs_scripts 02:10:41 (42), docker_containers 02:10:44 (11).
• 10 DB-fn sources recomputed at census time 02:10:50 UTC — COVERED (live recompute daily).
• kb_sop_docs — PARTIAL, last scan 2026-06-04 10:15 UTC (≈22h), status PARTIAL. It is NOT in the systemd pipeline (MCP/manual), so it ages until manually refreshed → labelled PARTIAL (and STALE_PARTIAL once >26h).

Rollup dashboard

instrumented_sources 16, covered 15, partial 1, stale 0, missing_adapter 0, unknown_classes 1, non_instrumented_classes 5, total_source_classes 21, instrumented_coverage_pct 94, verdict ADAPTER_COVERAGE_OK_WITH_DECLARED_PARTIALS.

The dashboard deliberately reports BOTH "16 instrumented" AND "21 total classes" so coverage is never overstated by treating the instrumented set as the whole universe.