Every number on this page is computed at request time from flock_slo_probes (5-minute synthetic probes) and flock_cron_runs (background-task failure log). Rolling 30-day window. No marketing math.
Process is alive and responding to requests.
Subscription checkout entry renders.
Supabase, Stripe, and Resend dependencies all reachable.
Event listing endpoint serves the discovery feed.
New-user signup landing renders.
Methodology
Synthetic probes hit five critical surfaces every 5 minutes via Vercel cron. Each probe records its HTTP status, latency, and a structured pass/fail classification. We expose the raw counts here so you can verify our uptime claims yourself.
"Degraded" weighs at 0.5 toward uptime — industry-standard for partial-availability events. "Down" counts as 0. 4xx user-error responses count as success per SLO.md.
Background-task failures are surfaced from flock_cron_runs, the cron-guard substrate introduced in Wave-Ultra-Review (2026-05-03).