AI agent safety checklist: sandbox, least privilege, human approval and logging

An AI agent that can only chat is harmless. The moment you give it tools, it can act: run shell commands, call APIs, send emails, delete files. And if an agent can act, sooner or later it will act wro