The gate for an agent belongs in the environment, not in the agent

There's a discussion running on Product Hunt right now about where an AI agent should stop and hand a decision back to a person. The framing that stuck with me: "can this be undone?" is the wrong gate...