Description
A session shows status: working and health.details: Claude is actively working even when CC has been stuck in a 'Misting...' state for over 1 hour with no output produced.
The monitor doesn't detect this as a stall because CC technically appears to be processing (the 'Misting' animation is running), but no text is being generated and the session is effectively dead.
Expected
The stall detection should catch sessions where CC has been in a single operation for an unreasonable amount of time (e.g. >15 min with no new output). The stallThresholdMs (300000ms = 5 min) should apply here.
Environment
- Aegis v2.2.2
- Claude Code v2.1.87 with GLM-5.1 model
- The session had spawned 2 sub-agents (both completed), then started 'Misting' indefinitely
Suspected Cause
The lastActivity timestamp may be getting updated by heartbeat/hook pings even when CC isn't producing output, preventing stall detection from triggering.
Description
A session shows
status: workingandhealth.details: Claude is actively workingeven when CC has been stuck in a 'Misting...' state for over 1 hour with no output produced.The monitor doesn't detect this as a stall because CC technically appears to be processing (the 'Misting' animation is running), but no text is being generated and the session is effectively dead.
Expected
The stall detection should catch sessions where CC has been in a single operation for an unreasonable amount of time (e.g. >15 min with no new output). The
stallThresholdMs(300000ms = 5 min) should apply here.Environment
Suspected Cause
The
lastActivitytimestamp may be getting updated by heartbeat/hook pings even when CC isn't producing output, preventing stall detection from triggering.