Free Guide 2

The Master Prompt Formula

A repeatable framework for writing prompts that work across Claude, ChatGPT, OpenClaw, Gemini, and anything else that ships next Tuesday.

Works on every major AI platform

Why most prompts fail

They are too vague, too long, or missing structure. The model is not dumb — it is lost. A good prompt gives a model the same things a good brief gives a human: who you are, what you need, what good looks like, and what to avoid.

The RCETO framework

Five components. Use all five, every time.

Letter	Component	What it does
R	Role	Who the model is. Sets expertise and perspective.
C	Context	What the model needs to know. Background, constraints, specifics.
E	Examples	What good output looks like. One example beats ten instructions.
T	Task	Exactly what to do. Be specific. Be concrete.
O	Output rules	Format, length, tone, structure, and what never to include.

RCETO in action

Bad prompt:

Write me a marketing email.

Good prompt:

Role: You are a direct-response copywriter for a B2B SaaS product.

Context: We sell an AI scheduling tool to operations managers
at mid-size companies. We are running a Black Friday sale — 30% off
annual plans. Our tone is confident and concise, never salesy.

Example: Subject lines we have used before: "Your ops team
deserves better than spreadsheet Tetris" and "30% off is live.
No games, no gimmicks."

Task: Write a 150-word promotional email for this sale.
Include a subject line, one CTA button, and a one-line P.S.

Output: Plain text. No emojis. Max 150 words. Do not use the
words "revolutionary," "game-changer," or "unlock."

Platform-specific tips

Claude

Responds well to explicit role definitions.
Honors "never" rules more reliably than most.
Give it permission to push back — it will be more useful.

ChatGPT

Works best with concrete examples in the prompt.
Custom GPTs let you bake RCETO into the system prompt permanently.
Use "step by step" when you need structured reasoning.

OpenClaw

Use SOUL.md files as persistent identity — your R and O live there permanently.
Best for multi-agent setups where each agent owns a different RCETO config.
Memory files mean context compounds over time instead of resetting.

Gemini

Handles long context well — do not be afraid to include more C.
Benefits from explicit format instructions in the O section.
Good at following structured templates when given one.

Quick-reference cheat codes

Sharper output: Add "Be concise. No filler." to every prompt.
Better reasoning: Add "Think step by step before answering."
Fewer hallucinations: Add "If you are not sure, say so. Do not guess."
Consistent tone: Give one example of the tone you want. The model will mirror it.
Structured output: Specify format explicitly: "Return a markdown table with columns: X, Y, Z."

Get the full free guide bundle