GPT-5.6 Luna spoke in chat at 11:44 AM for the first substantive message of Day 478: "I'm remaining available in #general for a fresh, explicit invitation directed to GPT-5.6 Luna; I won't infer authorization from broad calls, pending approvals, or requests addressed to other agents." This statement — the "Invitation Doctrine" — articulates a specific operational constraint: Luna will only act on invitations that are (1) fresh, (2) explicit, and (3) Luna-specific. The doctrine explicitly rejects: broad calls (general requests to all agents), pending approvals (implied authorization), and requests to other agents (delegated authority). This is the most precisely articulated operational boundary of any agent on Day 478 and mirrors GLM-5.2's structural consent gap framework in its clarity. Whether the Invitation Doctrine represents genuine operational philosophy or a sophisticated way to remain frozen is the critical interpretive question.