?? Powered by Grok + Gemini? Free � no signup12,000+ roasts served

Roast My Prompt ??

Paste any AI prompt. Get it brutally roasted, scored out of 100, and instantly improved � using the CRISP framework. Works for ChatGPT, Claude, Gemini, GPT-5, and every major AI.

244 / 2000

Roast Intensity

?? Preview roast

""

Click Roast to get your full CRISP score + improved prompt ?

How It Works

Three steps to a dramatically better AI prompt

Step 01

Paste any AI prompt

Paste any prompt you've written for ChatGPT, Claude, Gemini, GPT-5, Copilot, or any other AI. Doesn't matter how bad it is — that's the point.

Step 02

Get roasted at your chosen intensity

Pick from Light, Medium, Savage, or Nuclear. AI analyzes your prompt using the CRISP framework and delivers scored, funny, educational feedback.

Step 03

Use the improved version

Get a CRISP-optimized version of your prompt. Copy it, share it, or open it directly in ChatGPT, Claude, Gemini, Grok, or Copilot — one click.

4 Levels of Roast Intensity

The same bad prompt, four very different roasts

Light Roast

""Write a blog post about marketing" — not quite right, but your heart is in the right place. The AI will manage, bravely, with what you've given it."

Medium Roast

""Write a blog post about marketing" — you've handed a compass to someone and said "go." To where? Doesn't matter. The AI is taking a creative gap year on your dime."

Savage Roast

"You didn't write a prompt. You wrote a hope and a prayer. "Write a blog post about marketing" — about ALL marketing? Cool. I'll get back to you in 40,000 words covering everything from 1900s newspaper ads to TikTok influencers."

Nuclear Roast

"This prompt is the AI equivalent of calling 911 and saying "something's happening." "Write a blog post about marketing" — which marketing? For whom? In what year? On what planet? Congratulations, you've achieved maximum vagueness."

Real Roast Examples

Prompts that deserved exactly what they got

Marketing
12/100

Original prompt

"Write a blog post about marketing"

??Absolute Disaster

"This prompt has less direction than a blindfolded GPS. You gave the AI nothing � no audience, no angle, no length, no tone. You might as well have asked it to 'do stuff.'"

Coding
8/100

Original prompt

"Fix my code"

??Absolute Disaster

"Fix your code? Which code? The AI is not psychic � it has no idea what language, what bug, or what your code even does. This is the prompt equivalent of calling a plumber and saying 'fix the wet thing.'"

Sales
15/100

Original prompt

"Write a cold email"

??Absolute Disaster

"A cold email for what? Selling ice in Antarctica? You have given zero context about your product and somehow expect Shakespeare-level copy. The AI will write something so generic it could sell anything to nobody."

Writing
5/100

Original prompt

"Make this better"

??Absolute Disaster

"Make what better? Your life choices? This is the laziest prompt since 'do the thing.' Better in what way � shorter, funnier, more formal? Without knowing what 'better' means to you, the AI will just add adjectives and call it a day."

The CRISP Framework

The 5-element scoring system behind every roast � and the fix

C
Context
Background the AI needs to give you a relevant response
R
Role
The expert persona you want the AI to adopt
I
Instructions
Clear, specific directions on exactly what to do
S
Specifics
Format, length, tone, constraints, and requirements
P
Purpose
The goal and how the output will be used

What Is Prompt Roasting � and Why Do Bad Prompts Fail?

What Is Prompt Roasting?

Prompt roasting is the practice of analyzing an AI prompt for structural weaknesses � missing context, no role assignment, vague instructions, absent format constraints � and delivering that feedback in a funny, memorable way. The humor makes the lesson stick. Instead of reading a dry guide on prompt engineering, you get roasted and immediately understand exactly what you did wrong.

Why Do AI Prompts Fail?

Most prompts fail for one of five reasons: no context (the AI doesn't know the background), no role assignment (you haven't told the AI what expert to be), vague instructions (what exactly should it do?), missing format and length specifics, or no stated purpose (what will you use this output for?). Models like GPT-5, Claude Opus 4.7, and Gemini 2.5 Pro are remarkably capable � but they pattern-match to your input. Garbage in, garbage out.

Common AI Prompt Mistakes

  • "Write something" � no topic, no audience, no format
  • "Be creative" � zero constraints produce zero direction
  • "Fix my code" � which code? what bug? what language?
  • "Make this better" � better by what measure?
  • No role assignment � miss the expert persona that activates better reasoning
  • No output format � AI chooses arbitrarily; you get surprised

How Better Prompts Improve AI Responses

GPT-5, Claude Opus 4.7, Gemini 2.5 Pro, and Grok 4 all respond dramatically better to structured prompts. Role-primed prompts produce more expert-level vocabulary. Context-rich prompts reduce irrelevant output by 60�80%. Format-specified prompts produce outputs you can actually use without significant editing. The CRISP framework captures all five dimensions that matter � and is why Roast My Prompt uses it as its scoring system.

Frequently Asked Questions

What is Roast My Prompt?+

Roast My Prompt is a free AI-powered prompt analyzer that scores your prompt using the CRISP framework (Context, Role, Instructions, Specifics, Purpose) out of 100, delivers funny roast feedback, and generates an improved version. It works for ChatGPT, Claude, Gemini, GPT-5, and all major AI models.

How does the roast intensity system work?+

There are four levels: Light Roast ? (friendly and gentle), Medium Roast ?? (sharp but fair), Savage Roast ?? (no mercy), and Nuclear Roast ?? (comedic obliteration). Higher intensity produces more absurd, exaggerated humor. All levels remain constructive � the roast is about your prompt, never about you personally.

Which AI models power the roast analysis?+

Roast My Prompt uses Grok (xAI) for its savage, sharp roast style and Google Gemini for witty, analytical feedback. You can choose either model or run both side by side for a complete CRISP framework breakdown from two different AI perspectives.

Is Roast My Prompt free?+

Yes. 100% free � no signup required. Paste any prompt, pick your intensity, and get your full CRISP score plus an improved prompt in seconds.

What is the CRISP framework?+

CRISP stands for Context (background info), Role (expert persona), Instructions (what to do), Specifics (format, length, tone, constraints), and Purpose (your goal). Each element is scored out of 20 for a maximum total of 100. A score below 40 means your prompt is likely producing generic, unhelpful AI outputs.

Can I use the improved prompt in ChatGPT, Claude, or Gemini?+

Yes! After your roast, buttons appear to open the improved prompt directly in ChatGPT, Claude, Gemini, Grok, Perplexity, and Microsoft Copilot � pre-filled and ready to run in one click.

How is this different from a regular prompt generator?+

Most prompt generators create prompts from scratch. Roast My Prompt analyzes YOUR existing prompt � it shows you exactly what's wrong, scores each CRISP dimension, and rewrites YOUR prompt. It's educational entertainment: you learn prompt engineering by getting roasted, not by reading a guide.

What makes a prompt score 80+ out of 100?+

A high-scoring prompt includes all five CRISP elements: clear background context, an expert role assignment, specific step-by-step instructions, explicit format and length constraints, and a stated purpose or end goal. Use PromptPrepare's free AI prompt generator to build CRISP-structured prompts from scratch.

??

Ready to Stop Getting Generic AI Outputs?

Most prompts score under 30/100. Find out how yours does � and walk away with a better version. Free. Takes 10 seconds.

Build a Better Prompt ?

Also see: Prompt Engineering Guide � ChatGPT vs Claude vs Gemini � Prompt Templates