Exercise 03 — Tool-Calling Agent

Timebox: 45 minutes

Goal

Turn agent.py into a small tool-using agent that can call tools, stop cleanly, and explain its answer.

Work in

• agent.py

Tasks

1. Keep at least two tools working end-to-end.
2. Add an iteration cap so the agent cannot loop forever.
3. Log tool calls clearly.
4. Add a third tool or a better planner.
5. Upgrade the mock planner to a framework-driven or provider-driven version if you want the real interview shape.

Done when

• The agent calls tools when needed
• It stops with a final answer
• It does not spin forever on failures