What is a 'plan' in the agent loop?

• A. A list of all possible tools in the registry.
• B. A measure of agent performance.
• C. A static document unrelated to execution.
• D. A drafted sequence of steps and tool calls the agent intends to execute to achieve the user goal.

Answer: (D)

In the agent loop, a plan is the drafted sequence of steps and tool calls the agent intends to execute to achieve the user goal. It acts as a concrete route map guiding what to do next and in what order, based on the current context and available tools. The plan is dynamic: the agent can revise it as tool results come back, shifting steps, adding new actions, or abandoning others if new information changes the goal. It's not simply a registry of all possible tools, nor a performance metric, nor a static document; it's specifically about the intended execution path that drives interaction with the environment. For example, if the goal is to answer a question about a document, the plan might be to fetch the document, extract text, summarize, and then generate an answer, adapting as needed when a tool returns unexpected results.