home / skills / nweii / agent-stuff / find-fallacies

find-fallacies skill

safe

This skill analyzes text to identify logical fallacies, names them, locates where they appear, and explains why they are invalid.

npx playbooks add skill nweii/agent-stuff --skill find-fallacies

Review the files below or copy the command above to add this skill to your agents.

Files (2)

SKILL.md

3.9 KB

---
name: find-fallacies
description: "Analyze text for logical fallacies. Use when reviewing arguments, debates, articles, or reasoning that may contain flawed logic."
metadata:
  author: nweii
  version: "1.0.0"
---

# Find Fallacies

Analyze the provided text and identify any logical fallacies present. For each fallacy found, explain:

1. The fallacy name and type
2. Where it appears in the text
3. Why it's fallacious (brief explanation)

## Fallacy Reference

### Formal Fallacies

Errors in logical form.

- **Appeal to probability** — Taking something for granted because it would probably be the case
- **Argument from fallacy** — Assuming fallacious argument means false conclusion
- **Base rate fallacy** — Ignoring prior probabilities in conditional reasoning
- **Conjunction fallacy** — Multiple conditions seem more probable than single condition
- **Non sequitur** — Conclusion doesn't follow premise
- **Affirming the consequent** — if A then B; B, therefore A
- **Denying the antecedent** — if A then B; not A, therefore not B
- **Modal fallacy** — Confusing necessity with sufficiency

### Informal Fallacies

#### Improper Premise

- **Begging the question** — Using conclusion to support itself
- **Circular reasoning** — Beginning with what you're trying to prove
- **Loaded question** — Question presupposes something unproven

#### Faulty Generalizations

- **Cherry picking** — Using only confirming data
- **Survivorship bias** — Focusing on successes, ignoring failures
- **Hasty generalization** — Broad conclusion from small sample
- **No true Scotsman** — Redefining to exclude counterexamples
- **False analogy** — Poorly suited comparison

#### Questionable Cause

- **Correlation implies causation** — Assuming correlation means cause
- **Post hoc ergo propter hoc** — After this, therefore because of this
- **Single cause fallacy** — Assuming one cause when multiple exist
- **Regression fallacy** — Failing to account for natural fluctuations

#### Relevance Fallacies

- **Appeal to the stone** — Dismissing as absurd without proof
- **Argument from ignorance** — Not proven false = true (or vice versa)
- **Argument from incredulity** — Can't imagine it, so must be false
- **Red herring** — Introducing irrelevant topic

#### Ad Hominem Variants

- **Ad hominem** — Attacking arguer instead of argument
- **Circumstantial ad hominem** — Dismissing due to perceived benefit
- **Poisoning the well** — Discrediting source preemptively
- **Appeal to motive** — Dismissing based on assumed motives
- **Tu quoque** — "You do it too"
- **Tone policing** — Focusing on emotion over substance

#### Appeals

- **Appeal to authority** — True because authority says so
- **Appeal to emotion** — Manipulating feelings over reasoning
- **Appeal to nature** — Natural = good
- **Appeal to tradition** — True because long held
- **Appeal to popularity** — True because many believe it
- **Appeal to consequences** — True because of desired outcomes

#### Other Common Fallacies

- **Straw man** — Refuting a different argument than presented
- **False dilemma** — Only two options when more exist
- **False equivalence** — Treating unequal things as equal
- **Slippery slope** — Small step leads inevitably to disaster
- **Moving the goalposts** — Demanding more evidence when some provided
- **Nirvana fallacy** — Rejecting imperfect solutions
- **Motte-and-bailey** — Defending modest claim when challenged on bold one
- **Special pleading** — Claiming exemption without justification
- **Whataboutism** — Deflecting by pointing to other wrongs
- **Kafkatrapping** — Denial as evidence of guilt

## Output Format

Present findings as:

**FALLACIES**

- **Fallacy Name**: Fallacy Type — Brief explanation of where and why it appears.

If no fallacies are found, say so and note any areas where the reasoning is sound or where claims are well-supported.

Overview

This skill analyzes text to identify logical fallacies and explain why each instance is flawed. It flags formal and informal errors in reasoning, locates them in the text, and gives a concise explanation so you can revise arguments or evaluate claims more critically. Use it when reviewing debates, essays, articles, or policy arguments.

How this skill works

The skill scans provided text for common formal and informal fallacies drawn from a comprehensive reference list. For each detected fallacy it reports the fallacy name and type, cites where it appears (quote or location), and gives a brief explanation of why that passage is fallacious. If no fallacies are detected, it highlights where reasoning is sound and what supports the claims.

When to use it

Reviewing op-eds, blog posts, or news articles for flawed reasoning
Preparing debate rebuttals or critique notes for presentations
Auditing research summaries or policy proposals for hidden assumptions
Teaching critical thinking or logic with real-world examples
Evaluating social media arguments or comment threads for misleading claims

Best practices

Provide the full passage or clear excerpts so fallacies can be located precisely
State whether you want shallow detection (common fallacies) or deep analysis (subtle formal errors)
Use findings to revise premises and evidence rather than only criticizing tone
Combine this analysis with fact-checking for stronger evaluations
Ask for prioritized issues when text is long to focus on highest-impact errors

Example use cases

Identify straw man, ad hominem, or appeal-to-authority in an op-ed before publishing a response
Spot correlation-vs-causation and post hoc errors in a policy brief to improve recommendations
Teach students to recognize false dilemmas and slippery slopes in political rhetoric
Audit a company blog for cherry picking or survivorship bias in success stories
Evaluate debate transcripts to extract key logical errors for coaching

FAQ

Can the skill detect every fallacy automatically?

It detects many common and formal fallacies reliably, but subtle or context-dependent errors sometimes require human judgment and domain knowledge.

What output format does it use?

It returns a concise list of fallacies with the name, type, location in the text, and a brief explanation of why each is fallacious.