AgentOS adds a GET /metrics/refresh/status endpoint so clients can watch a background metrics refresh instead of waiting blind and timing out. It reports idle, running, completed, or failed along with started_at, finished_at, and any error, and the state updates even when a refresh finishes without writing new data, so a no-op still resolves cleanly rather than looking stuck. On the client it's available as AgentOSClient.get_metrics_refresh_status(), and for remote databases the status comes back from the remote AgentOS, so you get an accurate picture wherever your data lives.
View Agno’s Refresh Metrics docs to learn more.