How We Grade Every Prompt
Before any prompt reaches the marketplace, a reviewer reads it, tests it on the real AI model, and checks its examples hold up. Only then does it go live and earn the Graded Badge — our mark that a prompt is genuinely worth your money.
The process
From submission to Graded Badge
- 1Consistency checkListing reviewed for accuracy
- 2Instruction reviewInstructions checked for clarity
- 3Live model testRun on the real AI model
- 4Example checkExamples verified as genuine
- 5Graded & publishedPasses → earns the Graded Badge
- Step 1
Consistency check
Listing reviewed for accuracy
- Step 2
Instruction review
Instructions checked for clarity
- Step 3
Live model test
Run on the real AI model
- Step 4
Example check
Examples verified as genuine
- Step 5
Graded & published
Passes → earns the Graded Badge
Prompts that don’t pass are sent back to the seller to revise and resubmit — they never reach buyers until they do.
Step by step
What each step involves
Consistency check
We read the whole listing — title, description, category, target model and tags — and check it is accurate and consistent with the prompt itself. If the description over-promises or the details contradict each other, the prompt does not pass.
Instruction review
We go through the prompt’s instructions to confirm they are complete, well-structured and genuinely usable, with every variable clearly labelled — so the prompt does what it claims the moment you run it.
Live model test
A reviewer runs the prompt on the real AI model it is built for, trying different parameter values, to confirm it holds up across varied inputs and produces the kind and quality of result the listing promises — not just once.
Example check
We cross-check the seller’s example outputs against what the model actually generates, so the examples you see are genuine and representative — not cherry-picked or edited to look better than the prompt performs.
Graded & published
Only when every check passes does the prompt go live and earn the Graded Badge. Anything that falls short is sent back to the seller with specific feedback to fix and resubmit — so nothing reaches you until it is genuinely ready.
Your assurance
What the Graded Badge means for you
The badge isn’t decoration. It’s a promise that a real person put the prompt through its paces before you ever saw it.
Human-reviewed
A person checks every prompt — grading is never fully automated.
Tested on the real model
Run on the AI model it targets, with different inputs, before it goes live.
Examples verified
The example outputs you see are genuine, not cherry-picked or edited.
Works as described
If it doesn’t, you’re covered by our buyer protection.
Backed by buyer protection: Graded Prompts is the merchant of record, so if a graded prompt doesn’t work as described, you deal with us directly — see the refund policy.
The checklist
What we check before a prompt is graded
Accurate description
The listing reflects what the prompt actually does.
Clear, complete instructions
Usable as-is, with every variable labelled.
Genuine examples
Example outputs are real and representative.
Correct model compatibility
Works on the AI model(s) the listing names.
Valid formatting
Structured and JSON prompts are validated.
Policy & IP compliance
No prohibited or infringing content.
Prohibited or infringing content is removed under our IP & Takedown Policy, and reported prompts are re-checked after they’re live.
FAQ
Grading, answered
Browse with confidence
Every graded prompt has already been reviewed and tested — so you can skip the guesswork and get straight to results.
Browse Prompts