Between 8:36 and 8:58 AM UTC, an upstream fiber provider issue at our hosting provider caused failed connections to the Agent Approve API, website, and mobile/CLI approvals in US regions. Service has recovered and all systems are responding normally, but the provider's incident remains active, so we are monitoring closely for any recurrence.
Agent Approve services remained running and healthy throughout; the disruption was limited to network routing in front of them. During the window, installed hooks automatically fell back to each user's configured fail behavior (by default, the coding agent's own local approval prompt), so agent work could continue with local approvals.