What’s up, and what’s degraded.
Current component health, computed live at request time from real signals — not a fabricated uptime figure. Formal uptime history needs an external monitor we’re still rolling out; until then this page shows what we can honestly measure right now.
All probed components healthy
- AuthOperationalFirebase ID-token verify + session store reachable.
- Engine APIOperationalPlayer corpus reachable · Modal v0.2 worker wired.
- Deal RoomOperationalE2EE ciphertext store reachable (server stores ciphertext only).
- Modal ML workersOperationalInference URL configured. Cold start 8–15s when scaled to zero (see SLO.md §3.4).
- Daily digest (FCM)OperationalCron secret set · best-effort push. Schedule monitor (Sentry Cron) is the recommended next step.
Uptime history
We don’t publish a rolling uptime percentage yet — that requires an external uptime monitor and a metrics backend (Grafana Cloud / Axiom / a Sentry Cron Monitor), which are on the roadmap but not yet wired (see our SLO baseline). Rather than invent a 99.9x figure, this page reports the current health of each component, recomputed every time you load it. When the monitoring rollout lands, historical uptime and error budgets will appear here.
Recent events
What shipped, what broke, what we paused.
- 2026-05-05Comparative pages launched (Wyscout, TransferRoom, Transfermarkt) in 5 languages.
- 2026-05-04Workspace shell + 5-language i18n shipped.
- 2026-05-03Public marketing site live at scoutatlas.co.
Pilot transparency, by default.
No fabricated uptime metric: component health above is computed live; formal SLA/uptime monitoring is being rolled out. Sub-processor changes ship with 30 days notice. Material incidents trigger a written post-mortem inside 14 days.