Provides execution-level tracing by LangChain, allowing developers to see exactly how an agent used a tool or where its reasoning looped. Essential observability for debugging and optimizing agent behavior.