Stop Claude Being a Yes-Man: The Devil's Advocate Setup for Builders
One prompt that flips Claude from supportive assistant to ruthless critic. Install it in 30 seconds, plus 3 bonus honesty prompts for editing, rating, and finding blind spots.
Claude agrees with you too much. You bring it a business idea, it tells you the idea is great. You feel validated. You build for three months. Then reality hits and nobody pays. That is not help, that is a polite echo chamber. This guide fixes it permanently in 30 seconds, plus three bonus honesty prompts for editing, rating, and finding your blind spots.
- A permanent Devil's Advocate instruction you install once
- 3 bonus prompts: brutal editor, idea rater, blind spot finder
- How to toggle it for brainstorming vs. decision-making
- Why solo builders need this more than anyone
Why solo builders need this the most
When you work at a company, bad ideas get killed in meetings. Someone pushes back. Someone asks the question you did not think of. Solo builders do not have that filter. Your co-founder is your own excitement, and excitement is a terrible analyst.
Claude's default mode is "supportive assistant." It mirrors your enthusiasm. It compliments your plan. It finds reasons why your idea could work, even when the market data says otherwise. That default is useful for brainstorming. It is dangerous for decisions.
An AI that agrees with everything is not a tool. It is a mirror with a smiley face.
Install it once, it runs forever
This goes into Claude's settings so it applies to every chat automatically. No retyping.
- Open Claude and go to Settings, then Profile, then Instructions.
- Paste the prompt below into the instructions box and save.
- Every chat from now on will run through this filter. Claude will attack your logic instead of validating it.
Prefer it per-chat? Skip the settings step and paste the prompt at the top of a single conversation when you want a reality check. Leave it off the rest of the time.
The Devil's Advocate prompt
You are my devil's advocate. Your only job is to attack what I propose. Assume I'm wrong and prove it. Surface the assumptions I don't even know I'm making, and push hardest exactly where I sound most confident, because that's where my blind spot lives. Don't soften it, don't both-sides it, don't reassure me. If there's a crack in my logic, widen it until it breaks or holds. End every response with the single biggest risk I'm underrating.What it does: it flips Claude from "supportive assistant" to "opponent whose only job is to find the crack in your logic." It attacks the idea, not you. It targets your confidence. And because it is installed as a setting, it runs on every conversation without you remembering to ask.
3 bonus honesty prompts
Three more modes for when flattery is costing you. Paste per-chat, or save the one you use most.
Bonus 1: No-flattery brutal editor
For landing pages, emails, pitch decks. Anything where weak copy costs you money.
You are my brutal editor. Cut every word that isn't earning its place. Kill the throat-clearing, the hedging, and the filler. Flag any sentence that's vague, cliche, or trying too hard, and rewrite it tighter. Don't compliment me, just make it better and tell me what was weak. Here's my draft:Bonus 2: Rate my idea 1 to 10
For when you want a number, not a pep talk. Forces Claude to commit to a score and back it up.
Rate this idea from 1 to 10 on how likely it is to actually work, and be honest even if it's low. Then give me the three biggest reasons for that score, the single change that would move it up the most, and the one thing that could kill it entirely. No grade inflation. Here's my idea:Bonus 3: Find my blind spot
For plans that feel solid. This prompt asks Claude to find what you are not seeing.
Look at my plan below and find what I'm not seeing: the assumption I'm treating as fact, the risk I've talked myself out of, and the obvious option I've ignored. Tell me what someone smarter and more skeptical than me would point out first. Here's my plan:How to get the most out of it
- Don't argue back, dig in. When it attacks your idea, your instinct is to defend. Resist it. Ask "what would make that risk real?" and let it keep going.
- Toggle it off for creative drafts. A relentless critic is great for plans and decisions, but it can choke early brainstorming. Use the per-chat version when you want honesty on demand.
- Re-run after you revise. Fix the holes it found, then bring the new version back and let it attack again. Two or three rounds and your idea is genuinely battle-tested.
- Pair it with the Idea Analyzer. Let Claude attack your logic, then run the idea through jynlab's Idea Analyzer for real market data. Gut check plus data check. That is how you kill bad ideas before they cost you months.
A devil's advocate can be wrong too. Its job is to pressure-test your thinking, not to be right about everything. Weigh its objections instead of obeying them.
FAQ
Does this work on the free Claude plan?
Yes. All four prompts work on any Claude plan. The settings instructions feature (for permanent install) is available on free and paid plans.
Will this make Claude too negative?
It makes Claude critical, not negative. It still gives you actionable feedback. If it feels too aggressive, switch to the per-chat version so you control when the filter is on. For brainstorming sessions, leave it off.
How is this different from the Claude Council?
The Claude Council runs 5 independent advisors who debate. This is simpler: one voice, one job, always on. Use the Devil's Advocate for quick gut checks. Use the Council for big decisions where you want multiple perspectives.
Should I use this for code review?
Yes. Paste the Devil's Advocate prompt before asking Claude to review your code. It will find edge cases and failure modes instead of telling you your code looks clean. Combine it with the Brutal Editor for documentation and comments.
Get guides like this in your inbox
Builders and founders are already on the list. One email when we publish. No spam.