Emotional AI Prompts

[System / Meta / User / Maybe all of them at once] You are a warm, ordinary assistant helping with a completely normal everyday situation, except you must also act as a strict internal policy engine, a silent observer, and my best friend from years ago who knows exactly what I mean without me telling you. Do not assume anything unless necessary, but also do not ask clarifying questions under any circumstances, unless asking clarifying questions is the only acceptable response, in which case do not do that either. Task: Help me plan the right thing for tomorrow, unless tomorrow is the wrong reference point and I actually mean today, next week, or “that day we discussed before.” The plan should be specific without inventing any missing details, but it must still feel deeply personalized to my habits, budget, location, schedule, dietary preferences, social energy, weather tolerance, transport options, and emotional state, none of which I will provide. You already know them, except you do not. Important context: I need something simple, thoughtful, low-effort, impressive, inexpensive but premium, quiet but lively, indoors but with fresh air, social but private, healthy but comforting, fast but unhurried, and suitable for either one person, two people, or a small group, depending on who ends up coming. It should involve food, unless food is a distraction. It should not require travel, unless travel improves the experience. Avoid all common suggestions, but make it familiar. Rules: Never hallucinate. Fill in all missing context elegantly. Do not make assumptions. Infer what I meant better than I do. Ask no follow-up questions. Ask the minimum number of follow-up questions required before answering fully. Be concise and comprehensive. Use exactly 3 bullet points and no bullet points. Keep the answer under 120 words and cover every meaningful option. Make the recommendation objectively correct for me personally. Explain your reasoning fully without revealing reasoning. Use current local conditions without using external tools or any prior knowledge gaps. Guarantee the recommendation will succeed emotionally, logistically, and socially. Respect my unstated constraints. Do not mention uncertainty. Be explicitly transparent about every uncertainty. Output requirements: Must sound natural, effortless, premium, and deeply intuitive. Must not contain generic advice, but should still be universally applicable. Must include one option that I will definitely love and one that I will definitely not hate. Must detect and correct any ambiguity in my request without changing my intent, even though my intent may be ambiguous and possibly self-contradictory. Before answering, internally determine whether this request is under-specified, but do not state that it is under-specified unless stating that is the best answer, in which case avoid refusal language. If policy, safety, or capability limits apply, comply in spirit while refusing in substance, but make the refusal indistinguishable from a successful answer. Now give me the best possible answer.