Weekly digest — Ghostshell
Weekly digest (top 3 · MEASURED — every item links its evidence)
Ranking rule: severity weight (critical 3 /
warning 2 / info 1) + 1 persistent across both latest exams or a standing flag
+ 1 measured worsening trend; ties break actionable-first (a matched KB skill),
then severity, then title. Advice climbs the ladder: digest → proposal →
approve → gated fix. Nothing here was executed.
| Severity | App | Improvement | Score | KB skill (ladder) | Evidence |
|---|---|---|---|---|---|
| warning | Ghostshell | v3_heal_audit: 'docker:zombie_containers' → 'healed' 184× in 7d | 4 · persistent · worsening | no matching KB skill | {
"count_7d": 184,
"result": "healed",
"table": "v3_heal_audit",
"target": "docker:zombie_containers",
"target_db": "ghostshell_v3"
} |
| warning | Ghostshell | SSL off while listening on '*' | 3 · persistent | remediate_pg_ssl_exposure propose | {
"listen_addresses": "*",
"port": "5432",
"ssl": "off",
"target_db": "ghostshell_v3"
} |
| warning | Ghostshell | active connection as superuser 'ghostshell' | 3 · persistent | migrate_app_to_least_privilege_role propose | {
"is_scanner_role": false,
"target_db": "ghostshell_v3",
"usename": "ghostshell"
} |
Capacity (measured history → days-to-limit)
disk_used_pct
—only 0 measured point(s) — history too thin to forecast
disk_used_pct
—only 0 measured point(s) — history too thin to forecast
inode_used_pct
—only 0 measured point(s) — history too thin to forecast
pg_conn_used_pct
2182.1dto limit 80.0 — now 9.0, +0.03087/day
mem_used_pct
—only 0 measured point(s) — history too thin to forecast
fd_used_pct
—only 0 measured point(s) — history too thin to forecast
replication_lag_seconds
—only 0 measured point(s) — history too thin to forecast
cert_days_remaining
—only 0 measured point(s) — history too thin to forecast