Observed stale runs in /var/lib/omni/pipeline/scheduler-state.json: pipeline-omni-fund-web-hs-t-654-1771554726, pipeline-omni-agentd-hs-t-362-1771554727, pipeline-omni-agent-prompt-compile-hs-t-394-1771554728, pipeline-omni-agent-chat-hs-t-462-1771554730, pipeline-omni-agent-memory-hs-t-481-1771554731, pipeline-omni-agent-context-hs-t-485-1771554732, pipeline-omni-agent-interpreter-sequential-hs-t-486-1771554734, pipeline-omni-agent-hs-t-502-1771554735. agentd status reported running with 0 cost and no end_time, but there was no docker container and events had only workflow_loaded plus stderr text ('/root/.local/share/agent/auth.json ... permission denied'). Expected: status/list-runs should reconcile with container state or emit synthetic failure when startup aborts without terminal JSON event.
No activity yet.