Free AI Prompt & Custom GPT Checker
Give your AI the brief it's missing. Paste your custom GPT instructions or system prompt. Get a score and the top issues in 20 seconds.
What the checker looks at
Four dimensions, the same rubric Briefsy uses to audit full business context Packs.
Does it say who the assistant is, what it does, who it serves, how it should sound and the facts it needs? Missing pieces become guesses.
Could a reader act on every instruction? Vague lines like "be helpful" or "answer professionally" leave the AI to interpret.
Do any statements contradict each other — two prices, two policies, two sets of hours? The AI will pick one at random.
Is anything repeated or padded? Every duplicated line wastes the assistant's attention and can drift out of sync.
How the score works
Every issue the checker finds lowers one of the four sub-scores; the overall score blends them — 30% Completeness, 25% Clarity, 25% Consistency, 20% Efficiency. A contradiction costs far more than a duplicate line. One open critical issue caps the score at 69; two or more cap it at 59.
- 80–100Strong — reliable, specific, ready to use
- 60–79Fair — works, but the AI will guess at gaps
- 0–59Weak — expect vague or wrong answers
The difference specifics make
"Answer questions about our services and prices. Be friendly and professional."
Which services? Which prices? The AI invents both.
"We unclog drains ($149 flat), repair water heaters (from $220) and repipe older homes (quoted after a free visit). If a price isn't listed here, offer a callback — never estimate."
Exact services, exact prices, and a rule for everything else.
Who it's for
You wrote instructions for a GPT and want to know why it still hallucinates prices or policies.
You rely on ChatGPT, Claude or Gemini to answer customers and need those answers to be right.
You ship assistants for clients and want a fast quality gate before handing them over.
Frequently asked questions
Is it free?
Yes. You can check up to 3 prompts a day without an account. Creating a free account removes the limit and lets you fix the issues it finds.
Do you store my prompt?
Your prompt is stored privately so you can view and share your results. It is never published or used to train AI models, and it is deleted automatically after 7 days unless you save it to an account.
Which AI tools does it work with?
Any assistant that follows written instructions: Custom GPTs, ChatGPT Projects, Claude Projects, Gemini Gems, and raw system prompts for API-built assistants.
What's a good score?
80 or above is strong — the assistant has what it needs and rarely contradicts itself. 60 to 79 is fair: usable, but with gaps it will guess at. Below 60 is weak: expect vague or wrong answers.
How do I fix the issues it finds?
Create a free account and your prompt is imported as a Briefsy Pack: the same content split into editable blocks, with the issues listed as one-click fixes or questions to answer. A full auditor then re-scores it as you improve it.
What is context engineering?
Context engineering is the practice of giving an AI the right facts about your business — who you are, who you serve, what you offer, how you sound and your rules — so its answers are accurate instead of generic. A well-structured prompt is the simplest form of it.
Ready to check your prompt?
Twenty seconds, no account, nothing published.