With an understanding of agentic AI system components from the previous chapter, we now focus on the practical methods for directing these agents. This chapter provides techniques for precise agent control using advanced prompt engineering. You will learn to structure prompts for sequential operations, refine instructions for complex agent actions, and assign roles or personas to shape agent responses. We will examine how to use few-shot examples for agent guidance, apply reasoning frameworks like Chain-of-Thought ($CoT$) and Tree-of-Thought ($ToT$), and manage agent state through prompt design. Additionally, techniques for prompting agents for self-correction and error handling will be covered, culminating in practical exercises to develop coherent agent behavior.