Continual Policy Programming for LLM Agents: Learning Executable Policies from Online Experience

LLM agents can solve increasingly complex long-horizon tasks, while continually improving from ongoing interaction remains an important challenge. We introduce \emph{Continual Policy Programming}, a framework that enables a frozen LLM agent to learn from fine-grained online experience by updating a persistent executable policy. Policy updates take effect during the current task and persist across future tasks, allowing newly acquired knowledge to accumulate over time. Across ALFWorld, MiniGrid, and TextCraft, Continual Policy Programming enables agents to recover from failures online, reuse newly acquired behaviors across subsequent tasks, and accumulate capabilities over time. Our results show that persistent executable policies provide an effective way to turn streaming interaction experience into cumulative behavioral improvement.