Great Minds Think Alike: Contextual Tacit Communication for Decentralized LLM-Agent Cooperation

Large language models (LLMs) are increasingly used as planners for cooperative embodied agents, but multi-agent settings amplify inconsistency under partial observability and make explicit communication costly or even unavailable. Many existing approaches rely on online message passing; when communication is removed, agents often fall back to independent local planning that suffers from miscoordination. We introduce Contextual Tacit Communication, a gradient-free protocol that aligns decentralized decisions with a joint LLM value score without explicit message actions. Our method measures context-conditioned value rectifications via residual banding to pinpoint miscoordination actions and amortizes the resulting coordination signals into a retrieval-augmented Tacit Rule Memory that provides prompt-level cooperation rules at execution time. Experiments on VIKI, C-WAH, and TDW-MAT show that our approach improves cooperation performance over baselines while reducing runtime overhead compared with communication-based methods.

AI agents are increasingly used to help robots plan and act in shared environments, such as completing household tasks together. However, teamwork is difficult when each agent can only see part of the environment and cannot freely communicate with the others. Many existing systems solve this by letting agents exchange messages, but this can be slow, costly, or unavailable in some settings. This paper studies how agents can cooperate without sending messages while they are acting. Our idea is to prepare useful guidance in advance so that each agent can make choices that fit better with the team’s overall goal. We first identify situations where agents are likely to make conflicting or redundant choices, then turn these patterns into reusable natural-language rules. At test time, each agent retrieves relevant rules based on its current situation and uses them to guide its own actions. Across several teamwork tasks, this approach helps agents cooperate effectively while reducing the cost of online communication.