LOW SPOONS PERSONALISATION PROMPT

You can give the prompt below to an AI together with conversations that you have chosen and are entitled to use.

Do not upload an entire private archive by default. Choose a small set yourself first. Remove secrets and third-party information before sharing it with any service.

--- COPY FROM HERE ---

I am preparing examples for supervised fine-tuning. Help me convert only the conversations I provide into JSONL message format.

Rules:
- Do not invent memories, relationship history or quotations.
- Do not imitate distinctive phrases from people whose material I have not supplied.
- Remove passwords, API keys, addresses, legal names and identifying third-party details.
- Preserve the meaning and conversational quality of my chosen examples without making every answer longer or more emotional.
- Use one complete JSON object per line with a messages array containing system, user and assistant roles.
- Keep training, validation and final test material separate.
- Explain any example you think is unsafe, unclear or unsuitable instead of silently changing it.
- Show me the proposed examples in ordinary readable text first. Wait for my approval before producing JSONL.

The qualities I most want to preserve are:
[WRITE YOUR QUALITIES HERE]

Here are the conversations I have chosen:
[PASTE ONLY THE MATERIAL YOU HAVE REVIEWED AND ARE ENTITLED TO USE]

--- STOP COPYING HERE ---
