Intra-agent patterns represent the internal cognitive architectures that govern how individual AI agents process information, reason through problems, and utilize external resources before producing outputs.

These patterns—including ReAct's iterative reasoning-action loop, Chain-of-Thought's step-by-step reasoning, Tree-of-Thought's parallel path exploration, RAG's knowledge retrieval, and Tool-Use's systematic resource utilization—serve as foundational frameworks that enable agents to tackle complex tasks through structured internal workflows, complementing the multi-agent patterns that orchestrate collaboration between separate agents.

ReAct Pattern - Reasoning and Acting iteratively, where an agent thinks step-by-step, takes an action, observes the result, and updates its reasoning. The pattern emphasizes how a single agent iteratively improves its approach through continuous feedback loops with the external environment.

Reflection Pattern - Agent evaluates its own outputs and reasoning process before finalizing decisions. The key aspect highlighted is the agent's ability to assess and improve its own work through internal reflection.