ReAct Pattern
Agent action model: Reason, Act, Observe, Repeat.
Why it matters
ReAct gives agents a structured decision loop instead of acting blindly.
In practice
Claude Code follows ReAct: reasons about the task, acts (reads/writes files), observes results, plans next step.
Related terms
Chain of Thought
Prompting technique that encourages the AI to reason step-by-step before answering.
Planner
The agent component that breaks down high-level goals into executable steps.
Executor
The agent component that carries out planned actions by invoking tools, APIs, or workflows.
Agent (AI Agent)
An AI system that plans, decides, and acts autonomously toward goals, rather than simply responding to prompts.