When you share a plan with an AI, it tends to tell you it's a great plan. This skill is for when you want a better decision instead.
What it does
It works through your plan or decision in a fixed order:
- Assumptions: the one to three unstated beliefs the plan rests on, and what breaks if one is wrong.
- What's really being decided: sometimes "should I build this feature?" is really "am I avoiding talking to users?"
- Know vs hope: what you have evidence for, and what you're hoping is true.
- Weakest link: the single point most likely to make it fail, and why.
- The cheapest test: the smallest thing you could do this week to find out whether that link holds.
Then you get a short, direct read of the situation and one question that would most change the advice.
How it behaves
It skips the praise, but it doesn't invent problems either. If the plan is good, it says so in a line and moves on to what could still go wrong. If you push back with new information, it updates. If you push back without any, it holds its view politely.
Try it
- "Stress-test this plan:" followed by your plan
- "Be honest, what am I missing?"
- "I keep going back and forth on this decision"
Adapted from the Honest Thinking Partner idea in AI Skill of the Week, rewritten from scratch.
Install
In Claude Code, copy the folder into your skills directory:
git clone https://github.com/ecemkaraman/ai-skills.git
cp -r ai-skills/skills/thinking-partner ~/.claude/skills/
In the Claude app, zip the thinking-partner folder, then go to Customize → Skills, click +, choose Create skill → Upload a skill, and upload the zip. Code execution needs to be on in Settings → Capabilities.