Menu
Free AI skill based on Kahneman

Claude, laat jou Cognitive biases Scout before you decide

The AI Cognitive Bias Detector scans every decision, claim, quote, or plan for 30+ thinking errors from the work of Daniel Kahneman. You will receive concrete System 2 reframes and a debias action list within 60 seconds.

Based on Thinking, Fast and Slow (Kahneman, Nobel Prize 2002)
Detects 30+ documented biases from decision science
System 1 scores and concrete follow-up questions per bias
30+biases
6phase scan
The sixtiestot analyse
System 1 alert
6 biases found
Ready to reframe

Bias Scan: price increase 20%

System 1: 8 / 10
Overconfidence biasHigh
Survivorship biasHigh
Recency biasHigh
Confirmation biasmiddle
Anchoring biasmiddle
Debiasing actions5

Why a Bias detector Skill

You think you're making a rational decision. System 1 has already decided before System 2 is awake. This skill brings that difference to the surface before you make an irreversible mistake.

System 1 usually wins

Kahneman shows that quick intuition drives 95% of your decisions. This skill forces System 2 to the table.

30+ biases catalogued

From anchoring to survivorship bias. You get the name, the explanation, and the specific quote that proves the bias.

Concrete reframes

No vague advice. Each bias will receive a counter-question, an alternative scenario, and a measurable check.

Debias in 60 seconds

Paste your decision, get an immediate System 1 score, spotted biases and a debias action list for the next 24 hours.

How do Bias scan Works

Six phases in which Claude dissects your input. Not as a judge, but as a mirror. You see which cognitive biases are active and receive the questions that engage System 2.

Literal input

No summarising, no interpreting. The skill works with your words, including the definitive terms you use.

System 1 score

You will be given a score from 1 to 10, indicating how much gut feeling is in your reasoning, with a rationale.

Bias families

The skill runs along 6 families: availability, confirmation, overconfidence, social-affective, framing, and attention memory.

Proof by bias

Every spotted bias comes with a quote from your input. No bias without proof. No fabrication.

Concrete reframes

For the top 3 biases, you get a counter-question, an alternative scenario, and a measurable check to validate.

Action list 24 hours

Concluded with 3 to 5 concrete debias actions that you can implement immediately, from pre-mortem to base rate check.

1

Intake

Claude reads your decision, claim or plan. Flags definitive words such as “certainly”, “always”, “everyone” as red flags for System 1.

2

Diagnose

Classification: System 1 or System 2 reasoning. Score 1 to 10 with motivation in 3 sentences.

3

Bias scan

Systematically through 6 bias families. Minimum 3, maximum 7 biases with name, definition, and evidence.

4

Impact

Top 3 biases ranked by impact. Prediction of the most likely incorrect outcome if the biases remain.

5

Rephrase

For the top 3 biases: a counter-question that activates System 2, an alternative scenario, and a measurable check.

6

Action list

3 to 5 de-biasing actions within 24 hours plus final judgment in one sentence. Ready for action.

A real Bias scan in action

This is what you get when you ask Claude to scan a decision. Input: “We are going to increase our price by 20%, competitors did it too, this will definitely work.”

Bias Scan: 20 percent price increase

System 1 score: 8 / 10
What I hear

The user is considering a 20 percent price increase. Arguments: competitors have done the same with revenue growth, customer loyalty, no recent churn. Conclusion: “will definitely work”.

Active biases
  • 1
    Survivorship bias — “Competitors did it and revenue increased” High
  • 2
    Recency bias No churn in the last three months“ High
  • 3
    Overconfidence bias “This is sure to work” High
  • 4
    Confirmation bias — Only affirmative arguments mentioned middle
System 2 reframing for survivorship bias

Which competitors have increased prices and lost customers? Find three. Without that data, it's not a pattern but an anecdote.

Final verdict in one sentence.

This is a System 1 decision being presented as System 2, park for 48 hours and re-evaluate after a data check.

With and without Bias check

Why this difference matters for your next decision. It's not that you make fewer decisions. You make decisions with fewer blind spots.

without

Loose intuition

  • System 1 decides, System 2 rationalises afterwards
  • Only affirmative arguments are seen
  • Risks only become apparent when they become a reality.
  • Emotion and framing determine the outcome
  • No language to name one's own blind spot
  • The same mistake three quarters in a row
Met skill

Structured bias scan

  • System 2 comes to the table when you click “send”.
  • Biases get a name, a definition, and a quote
  • Prediction of the most likely incorrect outcome
  • Concrete counter-questions and measurable checks per bias
  • Action list to be completed within 24 hours
  • Learning speed doubles because you see your own patterns

This is the Skill Prompt under the bonnet

Transparent. The skill forces Claude through a fixed 6-stage process, leaving no room for easy reassurances. You see exactly what's happening as soon as you invoke the skill.

# skill: ai-cognitive-bias-detector # based on Kahneman 2011, Tversky Kahneman 1974
Role “bias detector in the tradition of Kahneman, not a judge but a mirror”
process Phase 1 — Intake: read literally, highlight assertive language as System 1 signal Fase 2 - Diagnose: Beoordeel Systeem 1 met 1-10 en een motivatie van maximaal 3 zinnen. fase 3 — bias scan: 6 families (availability, confirmation, overconfidence, social, framing, attention) Phase 4 — Impact: Top 3 ranked + prediction of adverse outcome Phase 5 — Reframes: Counter-question + Alternative + Measurable Check per Bias Phase 6 - Action list: 3 to 5 de-bias actions within 24 hours + final judgement in 1 sentence
rules Minimum 3, maximum 7 biases — quality over quantity Every bias requires a quote from the input as evidence Ensure there is no jargon without an explanation in a single sentence. maximum 1 clarifying question beforehand tone: calm, sharp, collegial — not patronising
output: “Large template with 7 header structure blocks” Sources “Kahneman 2011, Tversky Kahneman 1974, Morewedge et al 2015, Klein 2007”

Frequently Asked questions

The most incisive questions people ask before using the AI Cognitive Bias Detector.

Wat doet de AI Cognitieve Bias Detector precies?

The tool scans every text you provide (a decision, claim, plan, quote, argument) for cognitive biases. You'll receive a System 1 score, 3 to 7 spotted biases with quotes as evidence, concrete System 2 reframes, and a debias action list for the next 24 hours.

On which work is the skill based?

On Daniel Kahneman’s *Thinking, Fast and Slow* (2011) and his earlier work with Amos Tversky from 1974. Supplemented with debiasing research from Morewedge (2015), Larrick (2004), Klein (2007, pre-mortem), and Lord, Lepper, and Preston (1984, consider the opposite).

Which biases are detected?

More than 30, divided into 6 families: availability and representativeness, confirmation and hindsight bias, overconfidence and illusions of control, social-affective, framing and prospect theory, and attention and memory.

Is this skill free to use?

Yes, the skill is free to download and use within Claude. You place the SKILL.md file into your Claude skills folder and it is immediately available. MIT licence.

How do I install the skill in Claude?

Download the zip file using the button below. Unzip it and place the “ai-cognitive-bias-detector” folder in your Claude skills directory. Restart Claude. The skill will automatically trigger on terms such as “check for biases”, “am I thinking clearly” or "bias scan".

What is the difference with Pre-Mortem or Inversion Thinking?

Pre-Mortem and Inversion Thinker work with one technique. The Bias Detector is broader: it analyses each input for 30+ biases simultaneously and provides a structured report including a System 1 score. They are complementary.

Does this also work for personal decisions?

Yes. The skill is domain-independent. You can scan a business decision, but also a house purchase, a job application consideration, or a relationship choice.

What if the skill finds no biases?

Then he explicitly says that. Finding false positives is a bias in itself. The skill is instructed to only mention biases for which a quote from your input provides evidence.

Can I adapt this skill to my own sector?

Yes. The SKILL.md is an open text file. You can add bias families, replace examples, or adjust the output structure.

Install the Bias Detector in 2 minutes

Download the zip file, extract it into your Claude skills folder, and you're done. Free, MIT license, no account needed. Works on Claude Desktop and Claude Code.

SKILL.mdformat
MITlicence
No accountrequired
Claude 3.5+Compatible

Sea Claude skills from TheSEO

Think sharper by combining skills. Start with a bias scan, then move to a pre-mortem or inversion analysis for the sharpest decision.