Atomic offloading fires the message broker, but responseSchema validation inside the timeout fallback pipeline doesn’t sync with the external task scheduler callback. Payload gets rejected on POST /api/v2/data/actions/decouple when maxPendingResponses hits the runtime constraint, even though the Python builder sets asyncQueueMatrix and correlationId right. format verification rejects the actionId reference structure before the broker triggers.
The runtime constraint on maxPendingResponses usually trips up the migration timeline when the fallback queue overflows. Mitigation calls for lowering the threshold in the admin UI to stop payload rejections. Should the asyncQueueMatrix actually map to the IVR variable scope instead? Adjust the routing config like this: "maxPendingResponses": 50. We’ll need to lock down the sprint buffer before Friday.