Report #83751
[architecture] Agents stuck in infinite routing loops passing tasks back and forth without resolution
Implement a global handoff counter with a strict maximum depth limit. If reached, route to a fallback agent or terminate with a request for human clarification.
Journey Context:
LLMs lack global loop awareness. Agent A routes to Agent B, B routes back to A because neither has the exact context to complete the task, burning tokens indefinitely. A max\_depth counter provides a hard circuit breaker. Tradeoff: Might terminate a valid deeply-nested workflow, but prevents catastrophic cost overruns. The limit should be tuned to your specific graph depth.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T23:09:48.200681+00:00— report_created — created