Going live creates an audit schedule that is due immediately
The moment your first configuration deployment succeeds with acceleration enabled, a per-domain audit schedule is created (or reset) to run now — you do not have to remember to press "audit".
The trigger#
Inside the deploy flow, after the edge deployment is published, the platform calls scheduleFirstLiveDomainAudit, which upserts the domain's DomainAuditSchedule with nextRunAt: now (existing schedules get their clock reset to now too) and intervalHours from DOMAIN_AUDIT_SCHEDULE_DEFAULT_INTERVAL_HOURS — default 168 hours (7 days), configurable between 24 and 720.
The worker#
A scheduled worker claims up to 10 due schedules per cycle (enabled: true and nextRunAt <= now), advances each nextRunAt by the interval before running (so a slow audit never double-fires), and executes a comprehensive audit per schedule. A failure to schedule is logged and swallowed — deployment itself never fails because audit scheduling did.
The loop it starts#
First audit → readiness scores and the Your changes checklist become real (Open the checklist) → AI discovery report gains a sourceAuditRunId → llms.txt and content drafts are (re)evaluated. Re-auditing after you make changes is how the platform confirms the fix — every checklist item says so explicitly.