What problem does it solve? It answers whether a Personal AI / DSH infrastructure in OPERATIONS MODE is healthy, which issues need human attention, and which are only known external blockers, without silently mutating canonical state to make reports look green. ## Core Features & Use Cases - Multi-domain status aggregation: Runs existing checks for governance, durability (RPO/backup age), aic drift across harness targets, open proposals, and personalization behavior metrics, then classifies the overall result as NO ACTION, REVIEW, ACTION REQUIRED, or EXTERNAL BLOCKER. - Evidence freshness contract: Every domain reports both a status (HEALTHY/DEGRADED/BLOCKED/UNKNOWN) and an evidence state (CURRENT/LAST_KNOWN/UNAVAILABLE), so stale baselines are never passed off as live measurements. - Personalization behavior monitoring: Computes correction rate, repeat-correction sessions, over-personalization, and scope leakage from real DSH session messages, comparing against a calibration baseline report. - Use Case: Ask whether the system is healthy before an upgrade; the review distinguishes a real backup RPO breach from a known privacy-blocked repo push and only flags ACTION REQUIRED for genuinely new failures. ## Quick Start Use the personal-ai-operations-review skill to run a read-only health review of my Personal AI infrastructure and tell me whether anything needs action.
