LLM agents have rapidly evolved into autonomous systems, yet a persistent information gap remains between users and agents: communication is costly, while users' similar preferences further limit information exchange.
This paper formalizes Communication Policy, establishing textual and UI-based policies, and evaluates communication policies across diverse environments, personas, and model combinations.
To build information asymmetry for proactive agents, we set up two complementary settings, User-Agent and Planner-Executor.
Experimental results reveal complementary strengths between interaction channels: text-based interaction often facilitates task performance, while structured UI improves agents' response quality and persona compliance.
Motivated by that, a hybrid method combines these advantages.
We further propose Communication Policy Evolution (CPE), a self-evolution framework for refining communication policies through rollout and prompt-level evolution.
Without model modification, CPE achieves the best task success across multiple settings using prompt refinement alone.
Our findings identify communication behavior as a critical yet underexplored design dimension for LLM agents.
Blogger's Review: This paper delves into the communication strategies of LLM agents, introducing the innovative CPE framework that shows how optimizing communication methods can enhance agent performance. Future research could explore more complex interaction patterns to improve user experience and task completion capabilities.