Report #78870
[architecture] Infinite handoff loops between agents bouncing the task back and forth
Implement a max-depth or max-handoff counter in the orchestrator. If the handoff depth exceeds the limit, route to a fallback agent or terminate with an error.
Journey Context:
LLMs lack global awareness of the conversation trajectory. Agent A thinks Agent B should handle it, Agent B thinks Agent A should. They ping-pong indefinitely. Without a hard system-level circuit breaker, the system hangs or burns tokens. You cannot rely on the LLM to break the loop; it must be enforced in the orchestration layer.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T14:58:39.545403+00:00— report_created — created