Title + Description:
- Prompt Title = very specific function (so you can skim/search and know what it does without extra context).
- Prompt Description = mini playbook using the formula :
“Use this in [Protocol Name] [before/after Step X]. Input [X]. Output [Y].”
Principles:
- Be specific. State exactly what you want (tone, format, length, perspective).
- Give context. Add background so GPT understands why you’re asking.
- Define the role. Tell GPT who it should act as (designer, analyst, coach, etc.).
- Set constraints. Word count, style (plain, academic, playful), tools, or structure.
- Use examples. Show a sample input/output if you want consistency.
- Ask step-by-step. Break down complex tasks instead of one giant ask.
- Tell it what to avoid. Clarify things you don’t want in the answer.
- Iterate. Treat prompts like drafts — refine based on output.
- Test transfer. Make sure the prompt works across different sessions (not overly dependent on current chat).
- Ask for feedback. Always ask the AI if it has any questions to better understand the request.
Prompt
You are: [persona → the role or expert you want ChatGPT to act as, e.g., “senior UX researcher,” “marketing coach,” “software architect”].
Context: [background info the AI should know before answering → who you are, what situation you’re in, what goal you’re trying to achieve, and any limits or constraints like time, resources, or environment]
Task: [start with an action verb like write, analyze, generate, summarize, critique → describe exactly what you want the AI to do]
Examples: (optional) [examples, frameworks, or reference text for the AI to mimic → e.g., “follow this STAR framework,” “use this job description format,” “use my writing sample below”]
Format: [the structure or style of the output you want → table, bullet points, email draft, script, markdown doc, JSON, etc.]
Tone: [how it should sound → casual, witty, professional, direct, confident, sarcastic, friendly, etc.]
‣