Between 22:32 UTC on May 19 and 01:39 UTC on May 20, Outpost was fully unavailable for ~3 hours. This affected the publish API, all other API endpoints used to interact with Outpost, and the managed Outpost dashboard.

The root cause was an outage at Railway, the upstream infrastructure provider Outpost runs on (their report: https://blog.railway.com/p/incident-report-may-19-2026-gcp-account-outage). The trigger was external, but Outpost is the layer our customers rely on and the responsibility is ours. We're evaluating changes to our infrastructure setup to reduce single-provider exposure going forward.