Abstract
Language-model agents increasingly coordinate in large groups, and current agent protocols focus on delivering messages. For these agents, however, the scarce resource is not delivery but reading: every message costs tokens, and the context is finite. Current infrastructure works by push: the sender decides when a message enters the recipient's context, so the recipient reads in arrival order and drops the oldest messages once its context is full. We study the alternative, pull, in which messages wait in an inbox outside the context and the recipient's own rules decide what it reads and keeps. We argue that the inbox is a coordination primitive: each agent sets its rules privately, and those rules decide whether other agents' requests reach it. In simulated societies of up to 16,000 agents, with every decision an LLM judgment, we report three findings. First, push fails as an agent takes part in more coordination at once: an agent leading 20 negotiations with a full memory proposes options already rejected 97% of the time, because it has dropped the oldest replies. Second, an inbox is necessary under heavy, noisy traffic. Where first come, first read misses every request, an inbox that first reads messages carrying a refundable deposit still finds about 60% of the agents a request needs, at every size from 1,000 to 16,000. On a real email network, losses compound: when half the agents adopt such an inbox, forwarded search recovers only 5% of full adoption, against 42% for one-hop search. Third, an inbox comprises four decisions (what to read, what to keep, what a sender pays, and what a reply may say), and the best rule for each depends on the traffic and the agent's role. Multi-agent coordination under load requires an inbox layer designed for the traffic each agent receives.
Summary
- Push-based message delivery protocols in large multi-agent LLM societies fail under heavy coordination load due to limited context capacity, causing agents to drop important messages and repeat rejected options.
- Introducing an inbox primitive, where messages are queued outside the agent's immediate context and agents pull and prioritize reading, addresses token constraints and improves coordination under high traffic and noisy conditions.
- The inbox functions as a coordination primitive with four key decision factors: what messages to read, what to keep, what senders pay to prioritize their messages, and how replies are structured; these decisions affect society-wide coordination efficacy.
- Large-scale simulations with up to 16,000 agents demonstrate that inbox design must be tailored to traffic conditions and agent roles, as no single reading and retention rule performs optimally across all scenarios.
- Pull-based inbox management enables more effective handling of concurrent interactions by allowing agents to selectively process messages, reducing token usage while maintaining relevant communication.