Agents possess three core capabilities—autonomous planning, autonomous execution, and environmental interaction—which are the fundamental characteristics that distinguish Agents from ordinary LLM applications.