Practice — Case Study: Design a Coding Agent (5 questions)

You are designing a terminal coding agent that can read files, edit files and run shell commands in a developer's repository. A colleague proposes: "Give it one bash tool and put 'never run destructive commands' in the system prompt — the model is smart enough."

1. Explain why that is insufficient, using a concrete failure scenario.
2. Design the permission model: how you classify actions, what the defaults are, and which mechanisms enforce them.
3. How does the design change between an interactive session on a developer's laptop and an unattended run in CI?