The One Prompt Format That Works Across ChatGPT, Claude, and Gemini (And Why)

Over the past few months, I’ve tested hundreds of prompts across ChatGPT (GPT-5), Claude 3.5, and Gemini 1.5 Pro — from code debugging to essay writing, from API generation to creative brainstorming.

And while every model feels different — there’s one universal prompt format that consistently delivers 5× better answers across all three.

Let’s break it down.

The Problem with Most Prompts

Most people type prompts like this:

“Explain Java Streams with examples.”
“Write a Dockerfile for a Node.js app.”
“Give me a startup idea.”

That’s like walking into a library and shouting random questions at the librarian. You’ll get an answer — but not the best one.

Large Language Models (LLMs) don’t just need what you want.
They need context, role, format, and goal — to reason effectively.

The Universal Prompt Framework (WORKS Everywhere)

Here’s the structure I use — the RFGF Framework:

Role → Focus → Goal → Format

It looks like this

You are an expert [ROLE].
Focus on…