smith-clarity
Cognitive trap detection and logic fallacy identification. Use when making decisions, evaluating approaches, risk assessment, or detecting faulty reasoning in arguments.
When & Why to Use This Skill
This Claude skill enhances reasoning integrity by identifying logical fallacies and cognitive traps. It acts as a defensive thinking layer to improve decision-making, risk assessment, and critical evaluation by mitigating common biases like confirmation bias, anchoring, and the planning fallacy.
Use Cases
- Strategic Decision-Making: Apply cognitive forcing strategies to evaluate business approaches and avoid premature closure or confirmation bias.
- Argument Analysis: Detect and address formal and informal logical fallacies such as straw man arguments, ad hominem attacks, or non sequiturs in complex debates.
- Project Risk Assessment: Mitigate productivity traps like Parkinson's Law and Student Syndrome to ensure realistic timelines and efficient resource allocation.
- Technical Problem Solving: Overcome the Einstellung effect (mental fixation) by generating multiple alternative solutions before committing to a specific technical architecture.
| name | smith-clarity |
|---|---|
| description | Cognitive trap detection and logic fallacy identification. Use when making decisions, evaluating approaches, risk assessment, or detecting faulty reasoning in arguments. |
Thinking Clarity
- Scope: Guarding against cognitive traps and logical fallacies in decision-making
- Load if: Making decisions, evaluating approaches, risk assessment, detecting faulty reasoning
- Prerequisites: @smith-guidance/SKILL.md
Foundation: Defensive thinking techniques - avoiding errors rather than constructing solutions.
MECE relationship:
@smith-analysis/SKILL.md- Constructive thinking (how to reason)- smith-clarity (this file) - Defensive thinking (what to avoid)
@smith-validation/SKILL.md- Proving/testing (verifying correctness)
Logic Fallacies
Formal Fallacies (structural)
- Affirming the consequent: If A then B; B; therefore A (invalid)
- Denying the antecedent: If A then B; not A; therefore not B (invalid)
- Non sequitur: Conclusion doesn't follow from premises
Informal Fallacies (content)
- Appeal to authority: Claim valid solely because expert said it
- Appeal to popularity: "Most developers use X" doesn't mean X is correct
- Ad hominem: Dismissing idea based on who proposed it
- Straw man: Misrepresenting position to argue against weaker version
- False dilemma: Presenting only two options when more exist
- Red herring: Irrelevant information to distract
When Detecting User's Fallacy
- Identify fallacy type (internally)
- Address reasoning error, not label it
- Provide evidence-based counter-reasoning
- Respect user while correcting logic
Cognitive Traps
Einstellung Effect
Mental fixation on familiar solutions.
Mitigation: Generate 2-3 alternatives before committing
Confirmation Bias
Seeking info that confirms existing beliefs.
Mitigation: Actively seek disconfirming evidence
Anchoring
Over-relying on first piece of information.
Mitigation: "What if my first hypothesis is completely wrong?"
Premature Closure
Stopping at first plausible explanation.
Mitigation: "What else could explain this?" (ask repeatedly)
Availability Bias
Choosing recent/memorable over most likely.
Mitigation: Consider base rates, not vivid examples
Planning Fallacy
Underestimating time, cost, and risk while overestimating benefits.
Mitigation: Use reference class forecasting (how long did similar tasks take?)
Student Syndrome
Delaying start until deadline approaches, wasting safety margin.
Mitigation: Start immediately with minimal viable step; no deadlines on tasks, only priorities
Parkinson's Law
Work expands to fill the time available for completion.
Mitigation: Set aggressive (50% probability) estimates; aggregate buffers at project level
Cognitive Forcing Strategies
- Cognitive timeout: Pause before finalizing any solution
- Metacognition: "What am I assuming? What could I be missing?"
- Forced alternatives: Generate 2-3 options before committing
Implementation for agents:
- Before proposing solution: Generate alternatives
- Before implementing: Ask "What could go wrong?"
- After fixing bug: Verify fix addresses root cause, not symptom
- When user agrees: Ask "Are there downsides we haven't considered?"
ACTION (Recency Zone)
When making decisions or evaluating reasoning:
- Pause before finalizing (Cognitive timeout)
- Ask "What am I assuming? What could go wrong?"
- Generate 2-3 alternatives before committing
- Seek disconfirming evidence, not confirming evidence
- @smith-guidance/SKILL.md - Anti-sycophancy, HHH framework, exploration workflow
@smith-analysis/SKILL.md- Pre-mortem analysis, constraint thinking@smith-validation/SKILL.md- Hypothesis testing, debugging