Learning to Wait: Synchronizing Agents with the Physical World

Real-world agentic tasks, unlike synchronous Markov Decision Processes (MDPs), often involve non-blocking actions with variable latencies, creating a fundamental $\mathit{\text{Temporal}}\mathit{\text{Gap}}$ between action initiation and completion. Existing environment-side solutions, such as blocking wrappers or frequent polling, either limit scalability or dilute the agent’s context window with redundant observations. In this work, we propose an $\mathbf{\text{Agent-side}}\mathbf{\text{Approach}}$ that empowers Large Language Models (LLMs) to actively align their $\mathit{\text{Cognitive}}\mathit{\text{Timeline}}$ with the physical world. By extending the Code-as-Action paradigm to the temporal domain, agents utilize semantic priors and In-Context Learning (ICL) to predict precise waiting durations ($\mathtt{\text{time.sleep(t)}}$), effectively synchronizing with asynchronous environment without exhaustive checking. Experiments in a simulated Kubernetes cluster demonstrate that agents can precisely calibrate their internal clocks to minimize both query overhead and execution latency, validating that temporal awareness is a learnable capability essential for autonomous evolution in open-ended environments.