What the drift monitor last found for the domain, and what was done

Every live domain is checked against its records once an hour. This returns the last check (when, the result, the drifted records with expected and observed values), what was done about drift (put back automatically with the continuous access CustomDomain™ holds, or left for the tenant because the record was changed rather than removed, because there is no continuous access, or because automatic repair is not switched on), whether repairs are automatic for this domain and why, the records it is held to, and the drift history. An end user's widget token (scope connect) reads this only when it is bound to the domain and minted with the connection's own end_user_ref; use an API key otherwise.

Automatic repair runs only where the deployment switched it on (MONITOR_AUTO_REPAIR; repair.paused is true otherwise). It puts back only the record sets that disappeared, from the records CustomDomain™ wrote (a kept Domain Connect authorization re-applies its template, the only write it allows), and only for drift that began after a clean check. A record missing since monitoring started is left for the tenant. Repairs are paced and capped per hourly pass and per provider account, and after 3 repairs of a domain within 24 hours the monitor stops and reports fix_failed.