PROMPT ANALYSIS

The Prompt Autopsy.

Paste a prompt you've been using. We'll show you the gap between what you think you're asking for and what your prompt actually permits. Then we'll tighten it.

What does the Prompt Autopsy check?

✦
Missing context. Who it's for, what you sell, and what the output is for.
✦
No constraints. No length, format or tone, so the model falls back on the safest average.
✦
Ambiguous words. Words like “engaging” or “professional” that a model can read two ways.
✦
No examples. Nothing that shows the model what good looks like.
✦
Conflicting instructions. Two asks that pull against each other.
✦
No finish line. Nothing that tells the model when the answer is right.

How does it work?

01
Paste a prompt you actually use, up to 4,000 characters.
02
An AI model reads it the way another model would, without carrying out the task it describes.
03
You get what the prompt is trying to do, what it actually allows, each gap with a fix, and a tightened version to copy.

What does a result look like?

A real run on 1 October 2026 · trimmed, never edited

Pasted

Write a LinkedIn post about our new AI feature that gets lots of engagement. Make it professional but fun.

What you're asking for: Create a LinkedIn post announcing a new AI feature that balances professionalism with humor.
What it actually allows: Any post mentioning an AI feature in a professional-to-casual tone, regardless of whether it actually engages the audience.
Missing context: “our new AI feature” Without knowing what the feature does, the model generates vague platitudes about AI that apply to any product. Fix: Specify what the feature does, what problem it solves, and who benefits from it.
No finish line: “gets lots of engagement” The model cannot control engagement; this conflates intent with an unverifiable outcome. Fix: Replace with checkable criteria: post length, tone, structure (e.g., hook, benefit, CTA), or required elements.
Tightened prompt: Write a LinkedIn post (150–200 words) announcing [feature name], which [what it does and who benefits]. Tone: professional with one light, relevant joke or pun—no slang or emojis. Structure: hook (problem or insight), 2–3 sentences on the feature's benefit, one sentence on availability/next step, and a CTA (e.g., 'Try it free' or 'Learn more'). Include 3–5 relevant hashtags. Avoid hype words like 'revolutionary' or 'game-changing.' End with a question to prompt replies.

Five gaps in all; two are shown.

Questions

Is the Prompt Autopsy free?

Yes. No signup and no email. There's a daily limit per visitor so it stays free for everyone.

Does it run my prompt?

No. It analyses the prompt and never carries out the task inside it.

Do you keep the prompt I paste?

We don't store or log what you paste. It is sent to an AI model provider to be checked, under that provider's data policy, and the result comes straight back to your screen.

Which AI models is it for?

Any chat model. The gaps it finds, like missing context or no format, cost you output quality on all of them.

Also free: The Boring Marketing Detector. →

Built by MorningLabs AI — the AI lab inside The Morning Beer.