I think this answer could be slightly improved by making sure the agent understands what we can do with regard to adult content, creating a broad domain level blocking rule. It's clearly one of the most common requests so worth continuing to refine the clarity of messaging about what we are capable of in the system prompt