Create a personalized relationship prompt pack for my situation. Produce exactly 6 copy-paste prompts I can use with ChatGPT, Claude, or Gemini: prepare for a difficult conversation, state a boundary, repair after conflict, make a shared decision, understand a recurring pattern, and plan a weekly check-in. Relationship context: [RELATIONSHIP CONTEXT] Current situation: [CURRENT SITUATION] My needs and limits: [MY NEEDS AND LIMITS] Communication preferences: [COMMUNICATION PREFERENCES]
Safety concerns, if any: [SAFETY CONCERNS]
Write each prompt in first person and give it only the placeholders it needs. Make each one request a concrete output: a short conversation outline, a non-accusatory script, options with tradeoffs, reflection questions, or a check-in agenda. The prompts should distinguish observations from interpretations, name a feeling without assigning motive, make a specific request, and leave room for the other person’s autonomy.
For conflict prompts, include a pause option and prohibit mind-reading, scorekeeping, threats, diagnoses, or coercive language. For boundaries, require a statement of what I will do rather than an attempt to control the other person. For shared decisions, ask for criteria, options, and agreements to revisit the decision. Do not invent events, speak for the other person, or declare someone abusive based on limited information.
If there is fear, threats, stalking, violence, sexual coercion, control of money or movement, or immediate danger, prioritize a safety plan and appropriate local professional or emergency support over communication scripts.
Before answering, check that each prompt preserves consent and autonomy and that no advice relies on facts not provided. If a required input is missing, ask up to 3 clarifying questions; otherwise provide the 6 prompts only.Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [RELATIONSHIP CONTEXT] | Describe the relationship, its length, living arrangement, and relevant shared responsibilities. | Married for four years, living together, no children, both work full time. |
| [CURRENT SITUATION] | Describe the issue, recent events, and what you want to change using observable facts. | We have argued three times this month about one partner making weekend plans without checking first. |
| [MY NEEDS AND LIMITS] | List your needs, non-negotiables, and what you are prepared to do if a boundary is not respected. | I need shared plans discussed by Thursday; I will make my own plans if we cannot agree by then. |
| [COMMUNICATION PREFERENCES] | State preferences about timing, tone, written versus spoken communication, and conflict pauses. | Talk after dinner, no texting during work hours, take a 20-minute pause when voices rise. |
| [SAFETY CONCERNS] | State any concern about emotional, financial, sexual, or physical safety, or write none known. | None known. |
How to use
- Replace the context fields with observable events rather than labels such as selfish or uncaring.
- Choose one prompt for the immediate conversation instead of bringing six issues into it.
- Edit any suggested script until it sounds like something you would honestly say.
- Follow up with: “Make the boundary prompt warmer but keep the consequence clear and within my control.”
Variations
Boundary message
Use this when you need a brief, respectful statement of a limit.
Help me write a boundary message about [BEHAVIOR] for [RELATIONSHIP]. My need is [NEED], and the action I will take if this continues is [MY ACTION]. Give three versions: spoken, text message, and response to pushback. Use observable facts, an I-statement, a clear request, and a boundary I can enforce myself. Do not diagnose, threaten, or tell me to control the other person. Ask up to 3 questions only if a required input is missing. Check that the consequence is within my control.
Conflict repair
Use this after an argument when you want to take responsibility for your part.
Help me prepare a repair conversation after [CONFLICT]. My actions were [MY ACTIONS], the likely impact was [IMPACT], and what I want to do differently is [CHANGE]. Write a 150-word repair statement, five listening questions, and two specific follow-through commitments. Do not write an apology that demands forgiveness or shifts blame. Mark any assumption about the other person's feelings. Ask up to 3 questions only if a required input is missing.
Shared decision
Use this when partners, friends, or family members need to decide something with real tradeoffs.
Facilitate a shared decision about [DECISION] between [PEOPLE]. Our options are [OPTIONS], our constraints are [CONSTRAINTS], and our priorities are [PRIORITIES]. Produce a decision table with criteria, tradeoffs, unknowns, and reversible versus hard-to-reverse elements. Then write a 30-minute conversation agenda and a review date. Do not choose for us or invent preferences. Ask up to 3 questions only if a required input is missing. Check that each person's priorities are represented fairly.
Tips
- Write down what happened, when, and what was said before asking for a script; this prevents AI from turning a guess about motive into the center of the conversation.
- A boundary is strongest when it identifies your own response, such as leaving a conversation or changing access, rather than trying to compel behavior.
- Use a repair message only for the part you can own; an apology that also prosecutes the other person usually restarts the conflict.
- If safety is a concern, do not use a generated script as a test of whether someone will react safely; seek local support and prioritize immediate safety.
FAQ
Can AI tell me if my relationship is healthy?
It can help you name patterns and questions, but it sees only the information you provide. A qualified counselor or local support service is better for serious concerns.
Can I use an AI-written message word for word?
You can, but adjust it to your voice and facts. A message you cannot stand behind is unlikely to help.
What if the other person refuses to talk?
You cannot force a conversation. Focus on a clear request, your own boundary, and support for yourself if the issue continues.