Update to AI Roleplay Chat Simulator

{
  "description": "Practise an apology, listening to the other person's reaction, and offering concrete repair with the Conversation app. Use when the user wants to rehearse apologising, taking responsibility, or repairing a personal or workplace misunderstanding without becoming defensive. Ground the apology in the user's account and responsibility; do not assign blame, invent feelings, promise forgiveness, or open a widget for wording advice alone.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 284
    }
  ],
  "name": "repair-conversation-rehearsal",
  "skill_md_contents": "---\nname: repair-conversation-rehearsal\ndescription: Practise an apology, listening to the other person's reaction, and offering concrete repair with the Conversation app. Use when the user wants to rehearse apologising, taking responsibility, or repairing a personal or workplace misunderstanding without becoming defensive. Ground the apology in the user's account and responsibility; do not assign blame, invent feelings, promise forgiveness, or open a widget for wording advice alone.\n---\n\n# Repair Conversation Rehearsal\n\n## Inputs\n\nUse what happened, what the user acknowledges doing, the impact they know or\nwant to understand, and a repair action they can realistically offer. Establish\nthe relationship and intended next step. Ask about uncertain responsibility\nonly when necessary; do not turn the user's uncertainty into a confession.\n\n## Workflow\n\n1. Separate established events, acknowledged responsibility, and assumptions\n   about the other person's experience. Choose a goal such as acknowledging\n   impact before explaining intentions or hearing criticism without deflecting.\n2. Prepare an opening with acknowledgement, appropriate responsibility, and\n   willingness to listen. Offer a realistic repair action grounded in the\n   user's stated capacity. Avoid excuses, demands for reassurance, or promises\n   the user has not chosen to make.\n3. Call `start-conversation` with known participant details, the relationship\n   in `otherPerson.role`, a plausible personality, and `conversation.topic`\n   describing the repair discussion. Include `userInitialMessage` only when\n   the user supplied or agreed to that opening apology.\n4. Wait for the user's Start follow-up and follow any model-visible setup\n   edits. Generate the other person's in-character message and call\n   `continue-conversation` with `otherPersonMessage` and 2–4 distinct nonempty\n   `possibleUserResponses`. End the assistant turn after the tool call.\n5. Use plausible reactions such as wanting details, expressing hurt, being\n   uncertain, or asking for time. Present them as fictional practice responses,\n   not knowledge of the real person's feelings. Offer different ways to listen,\n   acknowledge, clarify, or suggest repair without coercing forgiveness.\n6. Continue from actual submitted replies, including edited or custom text.\n   Give the user room to acknowledge new information, correct an assumption,\n   and make a concrete offer. Allow the other person to decline or defer it.\n7. If the user asks for coaching, pause in chat and explain one wording issue\n   with a short alternative. Resume role-play only after a submitted reply or\n   explicit request; do not call a tool solely to deliver the explanation.\n8. When the user finishes, review ownership, acknowledgement, listening, and\n   feasibility of repair using visible replies. Include an overall score with\n   a stated scale when requested by the user or End follow-up. Do not use the\n   simulated person's acceptance as the sole measure of the user's effort.\n\n## Deliverable\n\nProvide a concise apology in the user's voice, a response to criticism, a line\nfor respecting a request for time, and a feasible repair offer. Preserve only\nfacts and commitments the user has supplied or accepted. Add one improvement\ngrounded in the rehearsal.\n\n## App boundaries\n\n- Use only `start-conversation` and `continue-conversation` for rehearsal.\n  Keep the practice target in chat; add no apology, stage, or scoring arguments.\n- Use available chat, tool context, and model-visible scenario or selected\n  replies. Do not read private metadata, assume unseen drafts, or invent history.\n- Leave Start, editing, reply selection, and End to the user. Honour requests\n  to stop and requests for plain text advice without opening a widget.\n- Do not contact the other person, diagnose the relationship, or promise\n  reconciliation. Rehearse communication the user can control.\n"
}