NOTICEPartial service degradation

Partial degradation.

Last checked May 28, 2026 · 09:57 UTC — updates posted within minutes of detection

5Services monitored
3Incidents · last 90 days
99.72%Overall uptime · 90 days
1Degraded right now
Preview environments is degradedBuild queue latency is above the normal threshold — new previews may take up to 4× longer to provision. Existing previews are unaffected. See the latest incident below for updates.
APIREST and GraphQL endpoints, auth, and rate-limiting
99.98%Operational
Agent computePer-session miniature server provisioning and task execution
99.94%Operational
Preview environmentsEphemeral build previews and live-reload tunnels
98.71%Degraded
DashboardWeb application and asset delivery
99.99%Operational
WebhooksOutbound event delivery and retry queue
99.97%Operational

The last 90 days, on the record.

Every incident, its root cause, and what changed afterwards. Post-mortems are linked within 48 hours of resolution.

Minor43 min

Preview environment build queues backed up

Preview environments
  • Monitoring alerts triggered on elevated P95 build latency. Engineering on-call paged.
  • Root cause identified: a deployment of the preview router introduced a queue-depth misconfiguration. Rollback initiated.
  • Rollback complete. Build queue draining normally. Latency returning to baseline.
ResolvedPost-mortem pending
Minor18 min

Elevated API error rates — EU region

API
  • Spike in 5xx responses detected on EU-West endpoints. Error rate peaked at 3.4%.
  • Cause traced to a stale connection pool after a scheduled database rotation. Pool forcibly recycled.
  • Error rate nominal. No data loss. Monitoring extended for 2 hours post-incident.
ResolvedPost-mortem pending
Minor29 min

Agent compute provisioning delay

Agent compute
  • VM provisioning times climbed above SLA threshold. Affected ~12% of new session requests.
  • Node capacity auto-scaling lagged due to a quota API timeout upstream. Manual scale-out triggered.
  • Provisioning latency back to normal. Quota API upstream confirmed healthy.
ResolvedPost-mortem pending

Showing the most recent 3 incidents. The full history is available on request — [email protected].

Get notified when something changes.

Status updates go out within 3 minutes of any state change — no polling needed.