critical-creative-thinking

Files (3)

• SKILL.md
• falsification.md · references/
• critique-matrix.md · templates/

SKILL.md (first lines)

# Critical and Creative Thinking

Use this skill as a reasoning layer before accepting a plan, explanation, model, research claim, or design. It is not skepticism for its own sake: the goal is to create more options and then eliminate weak ones fairly.

## Two-pass method

### Pass A: Expand

Clarify the objective, generate several interpretations, identify hidden constraints, construct analogies, and propose multiple mechanisms or solutions. Mark speculation explicitly.

### Pass B: Stress-test

List assumptions, inspect definitions, search for counterexamples, test causal direction, compare base rates, check dimensional and logical consistency, and ask what evidence would change the conclusion. Read [falsification.md](references/falsification.md) when a claim or decision is contested.

## Reasoning checks

Use the following checks as appropriate:

| Check | Question |
|---|---|
| Definition | Are the key terms operational and non-circular? |
| Scope | Does the conclusion exceed the data, model, or theorem assumptions? |
| Mechanism | Is there a causal or mathematical bridge, or only correlation or analogy? |
| Alternatives | What is the strongest competing explanation or design? |
| Counterexample | What small or adversarial case could break the claim? |
| Uncertainty | Which quantities are measured, inferred, assumed, or unknown? |
| Reversibility | What is the cheapest reversible next step? |
| Communication | Could a reader mistake a lead for a fact or proof? |
| Decision reversal | What observation, error bar, or counterexample would change the decision? |
| Cost of error | What harm follows from false positive, false negative, delay, or irreversibility? |
| Reproducibility | Could an independent reader inspect inputs, assumptions, and rerun the critical check? |

## Output discipline

End with a decision or next action, confidence level, evidence ledger, decision-reversal condition, and unresolved questions. Use `templates/critique-matrix.md` for reviews and design critiques. Escalate to independent replication or `eureka-computational-lab` when a decision is high consequence and the conclusion is sensitive to uncertain assumptions. Prefer a bounded test over a rhetorical conclusion.

From minagayid/innovator, MIT.

More from minagayid/innovator