Summary

The Reverse Brief flips the framing from "tell me what to do" to "hold these lines." You define what AI cannot touch, and Claude executes the work while respecting those boundaries. The output includes a flagged section showing where Claude would have crossed a line without your instruction.

When to Use

When drafting new work where you know certain moves, language, or claims would damage your positioning or misrepresent you. Use this before drafting so drift never makes it into the output.

Best Input

• A clear description of the task in normal terms ("Email to [Prospect]", "LinkedIn post about [Topic]")
• 3–5 specific boundaries phrased as "DO NOT [specific move]"
• Be concrete: Not "don't oversell" but "don't claim ROI numbers I can't back up"

The Prompt

I'm going to give you a task, but instead of telling you what to do, I'm going to tell you what NOT to touch. Treat these boundaries as hard stops.

Here's what I want the output to be about:

[DESCRIBE THE TASK IN NORMAL TERMS: "Email to [Prospect] about [Topic]", "LinkedIn post about my positioning", "Proposal response to [Company]", etc.]

Here are my BOUNDARIES — things you should never do, even if it seems like a good idea:

• DO NOT: [Specific move you want to prevent — e.g., "use motivational language", "claim expertise I don't have", "use the word 'innovative'", "suggest pricing options I haven't decided on"]
• DO NOT: [Another boundary]
• DO NOT: [Another boundary]

Now write the output respecting these boundaries.

After you write the output, add a separate section flagged as [BOUNDARIES PROTECTED]:

[List each boundary here and note whether you encountered a place where you would have crossed it if the instruction didn't exist. Example: "DO NOT use motivational language — I wanted to add 'This is a game-changer for your team' in paragraph 2 but held back per your boundary."]