Giovanni Pace
I'll refine the prompt guard rail by adding thematic constraints and detailed context, learning to keep the LLM model focused on the application domain without responding to off-topic requests.