Stop telling Claude what NOT to do in the prompts.

Here's the only prompt format that works on 4.7:

[Action verb] [exact task]. Length: [hard cap]. Format: [structure]. Tone: [3 words]. Go beyond the basics.

That's it. But you need to set it up first. Here's how:

Step 1. Open Claude. Select Opus 4.7 model.

Step 2. Turn on Adaptive thinking.

Step 3. Go to how-to-ai.guide.

Step 4. Subscribe for free. Open my welcome email.

Step 5. Hit the automatic reply button inside.

Step 6. Download ANTI AI WRITING STYLE .md file.

Step 7. Upload it to Claude. Use the format above. Add at the end of the prompt:

Think before answering (maximum reasoning).

After Claude's answer, follow up with this prompt:

Audit your text using the anti-ai-writing-style file. Rewrite anything that breaks the rules.

✦ Here's why your prompts don't work (red flags):

"Don't use jargon."

"Don't sound like AI."

"Don't use em dashes."

"Avoid buzzwords."

"Be conversational, not robotic."

These worked on Claude 4.6. They break on 4.7.

Your prompt says "don't" 14 times. Claude opus 4.7 reads each one literally.

The fix is counterintuitive.

_

Stop telling Claude 4.7 what to avoid.

Tell it exactly what you want.

"Length: 200 words" beats "keep it short."

"Tone: confident, specific" beats "don't be robotic."

"English a 16-year-old reads" beats fifteen "don'ts."

500-word negative prompt → robotic, literal output.

40-word positive prompt → as a human wrote it.

_

Get my full prompt guide: ruben.substack.com/p/pr…

♻️ Restack so others stop prompting like it's 2025.