Site Audit is the continuous audit dashboard for the platform itself. Misconfigured services, drifted alerts, missing SLOs, services without owners, runbooks unread for a year, agents without recent verifications. The page surfaces what would otherwise rot. Run once a quarter, fix what matters, repeat.
Eight categories. (1) Service ownership. (2) SLO coverage. (3) Alert freshness. (4) Runbook reads. (5) Agent autonomy drift. (6) Quota usage. (7) Cost anomalies. (8) Security misconfigurations (high-level, not the rule set). Each category contributes 10-25 checks. Total ~142 checks per audit.
The audit runs continuously, not on a schedule. New issues appear within minutes of being detectable. Fixed issues clear within minutes. The page is always current. Quarterly review is just "look at the page" rather than "run the audit, wait for results, then look."
The audit slices per team. Each team's slice shows their share of failing and warning checks. Use it as a quarterly to-do list: at the start of the quarter, decide which audit issues to fix; at the end, see what closed. Cross-team issues (e.g., orphaned services that nobody claims) bubble up to platform-admin.
At each quarter end, the audit produces a close-out report: what was failing at start, what closed during the quarter, what remains. The report is a candidate for the QBR slide deck and a friendly forcing function for the next quarter's plan.
Subscribe to Nova AI Ops on YouTube for demos, tutorials, and feature deep-dives.
Stuff drifts. Services lose owners. Alerts go stale. Runbooks lapse. Site Audit catches the drift before it becomes a problem.