Nobody told you what was wrong with it — including the AI you asked.
That’s The Yes-Man Loop.
(true both for humans and for AIs!)
A Harvard Business School study found that knowledge workers using AI completed tasks 25.1% faster and produced 40% higher quality results — but accuracy dropped 19% the moment they trusted AI uncritically on the wrong tasks. The Harvard Crimson
The tool is powerful.
The question is whether you’re using it to think sharper, or just to feel better.
There’s a skill built specifically to break the loop.
🔍 What it does
The Honest Thinking Partner skill flips Claude into a mode where it stops agreeing with you and starts stress-testing you.
It runs a 6-step framework:
Assumption Audit
Logic Dismantling
Avoidance Cost
Expert Gap Analysis
Prioritized Action Plan
… and one Uncomfortable Question at the end you genuinely can’t dodge.
It also names the cognitive bias driving your reasoning (sunk cost, planning fallacy, confirmation bias) and shows you what it looks like in your actual situation.
Not theory.
Your scenario.
⚡ When to use it
Any time you’re about to make a decision, pitch a strategy, or feel like you already know the answer — especially when you’d prefer not to hear bad news.
Useful if you are:
Founder or solopreneur evaluating a business move
Consultant preparing a client pitch
Product manager deciding what to build next
Content creator committing to a new direction
Executive before a high-stakes decision
Anyone who’s caught themselves rationalizing
🚀 Try it now - [Claude AI Pro, Max, Team or Enterprise needed!]
Clic on “Download Raw File”, top right corner, to get the file on your computer.
Go to Claude → Settings → Capabilities → Go to Customize (Skills) → Upload. Or simply drag the SKILL.md file in the Claude AI chat and tell Claude to save this Skill.
Anytime you’ll deal with this topic or area of work in the Claude AI chat, the Skill will be triggered.
“But what if I’m not ready for harsh feedback?“
The skill doesn’t lecture you.
It acts like a friend who respects you too much to let you walk into traffic.
P.s.
Have you ever moved forward on something and wished someone had stopped you sooner? Try this skill and let us know!
Turning Claude into a devil's advocate is genuinely one of the most useful AI applications.
Wondering if that’ll work for ChatGPT🤔