A user asks "What were the sales in 2023?" How do you prompt the LLM to separate the semantic search ("sales") from the metadata filter ("year == 2023")?

Agents generate a lot of "internal thought" and tool logs. How do you keep the context window from filling up with irrelevant logs during a long multi-step task?

Instead of the agent deciding one step at a time, why might you ask it to generate a full "Task List" first?

How do you evaluate an agent when the "correct path" might involve 5 different tool calls in any order?