Incidents and maintenance reported on status page for PromptGuard https://status.promptguard.co/ en https://status.promptguard.co/incident/1109094 Fri, 30 Oct 2026 21:13:18 -0000 https://status.promptguard.co/incident/1109094#dda825a093c3814402824d8829719c23fb4aa7ab198474962f65b83ed7c7d736 Incident Resolved at 08:22 UTC on 30 October 2026. This report is the postmortem, published after the fact. https://status.promptguard.co/incident/1109094 Fri, 30 Oct 2026 21:13:07 -0000 https://status.promptguard.co/incident/1109094#b2ff0858cb84a5f97d5b69db4f791dc9098e0d6084895794078eb8ad36da92bc Incident Between 20:41 UTC on 29 October and 08:22 UTC on 30 October, every request to the PromptGuard API failed, including Guard API scans and the Proxy. Requests were refused, not passed through unscanned. Capacity had been degraded since 13:05 UTC. **Cause.** A change to the files the API loads at startup, made in preparation for a release, took effect on the version already running. That version could not start with them, and as running instances were replaced, none came back. **Why it lasted 11 hours.** Our monitoring detected the outage within 13 minutes, but its alerts went to a channel nobody was watching overnight. **What we changed.** Startup files now change only together with the release that uses them. Alerts go to a monitored channel. This status page launched, with independent probes that run real scans every five minutes and publish outages automatically, and we now commit to a public postmortem within 72 hours of any customer-facing incident. We're sorry. If you were affected, reply to any PromptGuard email or write to support@promptguard.co.