SuperAPI AI Directory

logical-fallacy-detector

Agents & models

View repository

Updated 10-20 04:56

About this skill

Logical Fallacy Detector — Logical Fallacy Detection Tool

A specialized skill that enhances the judge and rapporteur agents' argument evaluation capabilities.

Target Agents

  • judge — Argument evaluation, logical fallacy identification
  • rapporteur — Include logic quality analysis in comprehensive reports

Major Logical Fallacy Classification

A. Relevance Fallacies

Fallacy Definition Debate Example How to Call It Out
Ad Hominem Attacking the person instead of the argument "Someone without experience has no valid opinion" "Please evaluate the argument itself"
Appeal to Authority Citing authorities outside their expertise "A famous actor said so too" "Do they have expertise in this field?"
Appeal to Popularity Majority opinion = correct "80% agree, so it must be right" "Does majority opinion equal validity?"
Tu Quoque "You do it too" as rebuttal "You have the same problem" "Your behavior does not invalidate the argument"
Straw Man Distorting the opponent's claim to rebut it "The opponent wants to eliminate all regulation" "Please quote the original claim accurately"

B. Causal Fallacies

Fallacy Definition Example How to Call It Out
Post Hoc Temporal sequence = causation "Crime dropped after the policy -> policy caused it" "What about other variables?"
Correlation = Causation Mistaking correlation for causation "Ice cream sales up -> crime up" "Have you considered a third variable (weather)?"
Slippery Slope Assuming extreme consequences "Allowing this will inevitably lead to ~" "Please prove the probability of each step"

C. Structural Fallacies

Fallacy Definition Example How to Call It Out
False Dilemma Limiting options to only 2 "Either for or against, nothing else" "Isn't there a third alternative?"
Circular Reasoning Premise presupposes the conclusion "This is right because it is right" "Please provide independent evidence"
Hasty Generalization Generalizing from insufficient cases "One case was like this, so all are" "Is the sample sufficient?"
False Analogy Comparing fundamentally dissimilar things "A national economy is like a household budget" "What are the key differences between the two?"
Red Herring Diverting to an unrelated topic Presenting an emotional case unrelated to the topic "Let's return to the original point"

D. Evidence Fallacies

Fallacy Definition Example
Cherry Picking Selecting only favorable data Citing only success cases from certain countries
Appeal to Ignorance No disproof = true "There's no counter-evidence, so it must be true"
Inadequate Sample Citing biased samples Surveys with only voluntary participants

Judge Evaluation Rubric (Logic Quality)

Argument Soundness Assessment Matrix

Criterion 1-20 pts 21-40 pts 41-60 pts 61-80 pts 81-100 pts
Logical consistency Multiple fallacies 3-4 fallacies 1-2 fallacies Generally sound Airtight logic
Evidence quality No evidence Weak evidence Standard evidence Quality evidence Academic evidence
Rebuttal effectiveness No rebuttal Superficial rebuttal Partial rebuttal Effective rebuttal Complete dismantling
Staying on point Frequent digressions Occasional digressions Generally on topic Mostly on topic Perfectly on topic

Fallacy Penalty Standards

Severity Penalty Example
Minor -2 pts Slight exaggeration, unclear source
Moderate -5 pts Hasty generalization, weak analogy
Serious -10 pts Straw man attack, false dilemma
Critical -15 pts Circular reasoning, data manipulation

Fallacy Identification Expression Guide (for judge)

Point out accurately while maintaining respect:

"In the [side] debater's [Nth] argument, a [fallacy name] fallacy is observed.
 Specifically, [quotation]; this constitutes [explanation of why it is fallacious].
 To improve this, [direction for improvement] would be needed."