Objective, Role, Action, Context, Layout/Length, Example — A well-rounded, six-part structure for everyday prompts.
ORACLE is the default choice when you're not sure which framework to reach for. It covers the six things almost any prompt benefits from: what you want, who the AI should act as, what to actually do, the background it needs, how the output should be shaped, and a concrete example. Use it for blog intros, emails, explainers, summaries, and general writing or research tasks.
The end result you're after — the single sentence that describes success.
The persona or expertise you want the AI to adopt while answering.
The specific task the AI needs to perform, stated as a clear instruction.
Background details, audience, or constraints the AI needs to get it right.
The format, structure, and rough length of the output.
A sample of the tone, style, or structure you want it to match.
Bad prompt:
“Write about email marketing.”
Structured with ORACLE:
Objective: Get more small business owners to open our weekly newsletter. Role: You are an email marketing strategist with experience writing for busy small business owners. Action: Write three subject line options for our weekly newsletter. Context: The newsletter covers one practical marketing tip per week. Our audience is time-poor and skeptical of hype. Layout/Length: Three subject lines, each under 60 characters, no emojis. Example: Similar in tone to "The 10-minute fix for your slowest page" — specific and low-hype.
Three subject lines matching the brief, each specific rather than generic, each under the character limit, with a short one-line note on why each one works for a skeptical, time-poor reader.
ORACLE's plain, labeled-section format works well across ChatGPT, Claude, and Gemini — it doesn't rely on any model-specific syntax, which is why PromptGeneratorLab uses it as the default framework.