send me your prompt

send me your prompt, not just your slop

Modeled off of https://nohello.net/en/ write some prose for sendmeyourprompt.com.<br>It's meant to be a page to show people who send me slop AI docs. Self-referentially include and explain<br>this exact message in the beginning of the page to show the pattern.

If you're reading this because I sent you sendmeyourprompt.com , there's a good chance you just<br>did something like this.

That message is a perfect example of what I'm asking for.

It contains the actual instructions that produced the output. It tells me the intent, the reference,<br>the audience, and the desired style. That's useful. Truly. 🙂

What is less useful is forwarding me only the generated document and expecting me to reverse-engineer the<br>conversation that created it.

AI output is a build artifact.

The prompt is the source code.

If you send me the compiled binary but not the source, I'm left guessing:

What was the original goal?

Which parts are intentional?

Which parts did the model invent?

What constraints did you give it?

What did you edit afterward?

Is this the first draft or the tenth?

Without the prompt, every change starts from archaeology.

With the prompt, I can reproduce, improve, critique, or regenerate the work in minutes.

"But the document is what matters."

Sometimes.

If you're sending me something that's final and you're asking for feedback on the content itself, the document<br>is enough.

But if you're asking me to:

improve it,

extend it,

adapt it,

debug it,

explain why the AI did something strange,

or continue the workflow,

then the prompt is usually more valuable than the output.

Prompts preserve intent.

AI text often loses the reasoning behind why it looks the way it does.

The prompt captures things like:

who the audience is,

what tradeoffs were requested,

which examples were excluded,

what tone was intended,

and what the model was optimizing for.

Those are impossible to reliably infer from the finished text.

Don't make me prompt-forensics your document.

Tim<br>2:15 PM

Did you ask for this section?

Tim<br>2:16 PM

Was this hallucinated?

Tim<br>2:17 PM

What problem were you actually trying to solve?

Every AI-generated document without its prompt turns into a detective story.

I end up asking questions like:

"Did you ask for this section?"

"Was this hallucinated?"

"Did you want bullets?"

"Was this supposed to be funny?"

"Is this copied from another draft?"

"What problem were you actually trying to solve?"

We could have skipped all of that by starting with the prompt.

Good

Dawn<br>2:15 PM

Here's the prompt I used.

Dawn<br>2:15 PM

Here's the output.

Dawn<br>2:16 PM

Here's what I don't like.

This gives me the source, the build artifact, and the part you want help with.

Even Better

Dawn<br>2:18 PM

Here's the chat link.

Tim<br>2:18 PM

Perfect. I can see the whole workflow.

Even better: send the chat link.

Less Good

Keith<br>2:15 PM

[15-page AI-generated document attached]

Keith<br>2:15 PM

Thoughts?

Less good: a long AI document with no prompt and no explanation of what kind of help you want.

Worst

Keith<br>2:20 PM

Can you fix this?

Tim<br>2:21 PM

Can you send the prompt too? 🙂

Worst: "Can you fix this?"

...with no context whatsoever.

I don't need perfect prompts.

Messy prompts are fine.

Half-finished prompts are fine.

Embarrassing prompts are fine.

A stream-of-consciousness dump is often better than a polished version because it tells me what you<br>were actually thinking.

I just want the source.

The Rule

If an AI wrote it...

Send me the prompt.

Not because I want to judge it. I promise. 🙂

Because that's where the information is.

The document is what the AI said.

The prompt is what you said.