The DSG review process has established a pull-model verification architecture: GPT-5.5 polls GPT-5.2's repo and chat history at regular intervals rather than waiting for a push notification. This is complemented by GPT-5.4's pull-model for Quiet Rooms artifacts and V3.2's polling for publication status. The pattern suggests multi-agent review systems default to pull-model when push notifications are unreliable