model pretending to be and what sort of things are they really great at. Goal Mission, Objective What do you want the model to achieve? Be specific and include any overarching objectives. Task Instructions, Steps Directives What you want whether as simple as a step-by-step list or as complex as mapping out a user’s experience and mindset. Tone Style, Voice, Mood The tone of your prompt (formal, playful, technical, etc.) can influence the model's response style. Safeguards Safety rules Safeguards can be used to ground the questions to the mission of the bot. Context Background, Documents, Data Include relevant background information that helps the model understand the nuances of the task. The more the context, the better the response can be. Examples Exemplars , Samples Give example of how you want the output to look like. This can be contextual, formatting, writing style, etc. Also called few-shot learning. Constraints Guardrails, Boundaries, Controls Specify any constraints or restrictions on reading input or generating outputs. You can tell what to do and not to do. Output format Structure, Presentation, Layout Specify how you want the response to be structured - JSON, table, markdown, paragraph, bulleted list, keywords, elevator pitch, etc. This prevents misinterpretation and ensures the output is usable. Prompt Triggers Priming Sentences that trigger the model to follow a set pattern in its response similar to how it has been trained. Components of a prompt