What makes Logical Fallacies notable?

1.1K downloads. Trust evidence available

How should Logical Fallacies be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Logical Fallacies

Activate when: someone says 'this argument feels off but I can't explain why', 'is this a real argument or just rhetoric?', 'what's wrong with this reasoning... Skill: Logical Fallacies Owner: deciqai Summary: Activate when: someone says 'this argument feels off but I can't explain why', 'is this a real argument or just rhetoric?', 'what's wrong with this reasoning... Tags: latest:1.0.5 Version history: v1.0.5 | 2026-07-16T18:05:10.879Z | user Description tail link + agents machine-readable metadata line (deciqai.com/s/logical-fallacies.json) v1.0.4 | 2026-07-10T10:26:53.800

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.1k

Updated

Oct 11, 2026

Version

1.0.5

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.1K downloads reported by the source. Last updated 10/11/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 11, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 11, 2026
Adoption signal: 1.1K downloadsadoption · observed Oct 11, 2026
Latest release: 1.0.5release · observed Jul 16, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s17a4mqcnk515kvaca5ze55d0x88pfpx:logical-fallacies

Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-deciqai-logical-fallacies/snapshot"

Documentation

CLAWHUB

137,737 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: logical-fallacies
description: "Activate when: someone says 'this argument feels off but I can't explain why', 'is this a real argument or just rhetoric?', 'what's wrong with this reasoning?', an argument relies entirely on authority/emotion/popularity, or you're about to decide based on a single analogy. Do NOT activate when: the conclusion is already verifiable empirically (just check the data); casual conversation where rigor is socially expensive and stakes are low. More: deciqai.com/c/logical-fallacies"
---

# Logical Fallacies

## Overview

A fallacy is an argument that looks like it works but doesn't. The test is not whether the conclusion is true — it's whether the *inference* from premises to conclusion is valid. This skill covers two layers: the **classical taxonomy** (Aristotle's 13, c. 350 BCE — verbal and structural errors) and the **modern cognitive map** (Tversky-Kahneman 1983 — errors competent reasoners commit automatically before any sophist arrives).

Composes with neighbors: `critical-thinking` audits evidence quality and framing; `first-principles` attacks premises; `mece` catches decomposition errors that masquerade as false-dichotomy or composition fallacies.

## When to Use

- An argument feels persuasive but you cannot articulate why
- A claim is supported entirely by authority, popularity, emotion, or anecdote
- You're about to decide based on a single argument or analogy
- A debate is moving fast ("everyone knows Y") — speed is the sophist's friend
- You catch yourself reasoning emotionally ("this has to be true because…")
- You're weighing an AI hype or AI-adoption claim ("a lab CEO said it's near," "it passed the benchmark so it's intelligent," "doom vs. utopia")

**When NOT to use:** casual small talk with low stakes; conclusion is empirically verifiable (just check the data); you're tempted to name a fallacy to dismiss an opponent rather than find truth (that is itself the fallacy fallacy).

## Coaching Novices (Adaptive Front Door)

- **Engine mode:** user has a concrete argument → run The Process directly.
- **Coach mode:** user signals unfamiliarity or has no concrete case → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. One-line what-it-is: some arguments *sound* right but don't earn their conclusion — this is a checklist for finding that gap, including in your own thinking.
2. Check fit against When to Use / When NOT to use. If data can answer it, say so.
3. Elicit their real argument — ask for a concrete case (something someone said, an article they're suspicious of). > **[WAIT — do not advance until user responds]**
4. Walk through the Audit one pass per turn: pose the question, wait for their answer, surface what they missed. > **[WAIT — do not advance until user responds]**
5. Close by naming the one fallacy *they* found and what changes about the conclusion now that they've seen it. > **[WAIT — do not a

_meta.json

{
  "ownerId": "kn754b8sk22s8c6gjxt02bftbn88q7ye",
  "slug": "logical-fallacies",
  "version": "1.0.5",
  "publishedAt": 1784225110879
}

references/sources.md

# Sources — logical-fallacies

> *Primary sources for the [logical-fallacies](../SKILL.md) skill.*

- Aristotle, *Sophistical Refutations* (*Peri Sophistikōn Elenchōn*), c. 350 BCE, trans. W. A. Pickard-Cambridge, in *The Complete Works of Aristotle*, Princeton/Bollingen. The founding taxonomy: six linguistic fallacies (equivocation, amphiboly, composition, division, accent, figure of speech) and seven non-linguistic (accident, hasty generalization, ignoratio elenchi, begging the question, false cause, complex question, affirming the consequent). Full text: https://classics.mit.edu/Aristotle/sophist_refut.html
- Stanford Encyclopedia of Philosophy, *Aristotle's Logic*, §3 "The Subject of Logic: Syllogisms" and §6 "Demonstrations and Demonstrative Sciences" — for the relationship between fallacy taxonomy and Aristotle's broader logical project. https://plato.stanford.edu/entries/aristotle-logic/
- Tversky, A., & Kahneman, D. (1983). "Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment." *Psychological Review*, 90(4), 293–315. The Linda problem; the empirical demonstration that competent reasoners (including statistically-trained subjects) systematically commit the conjunction fallacy by substituting representativeness for probability. https://doi.org/10.1037/0033-295X.90.4.293
- Tversky, A., & Kahneman, D. (1974). "Judgment under Uncertainty: Heuristics and Biases." *Science*, 185(4157), 1124–1131. The earlier programmatic paper introducing representativeness, availability, and anchoring as the cognitive substrate from which many modern fallacies arise. https://doi.org/10.1126/science.185.4157.1124
- Stanford Encyclopedia of Philosophy, *Fallacies* — a comprehensive modern survey distinguishing formal vs informal fallacies and tracing each through the literature from Aristotle onward. https://plato.stanford.edu/entries/fallacies/
- Walton, Douglas (2008). *Informal Logic: A Pragmatic Approach* (2nd ed.), Cambridge University Press — the standard modern reference on informal fallacy theory, treating fallacies as misuses of otherwise-legitimate argumentation schemes (e.g., when *ad verecundiam* is and is not legitimate).
- Stanford Institute for Human-Centered AI (HAI), *AI Index Report* (annual, 2024 edition and later). Documents the measurement problems behind AI capability claims — benchmark saturation and data contamination (test items leaking into training data), which is what makes "it passed the benchmark, therefore it's intelligent" an instance of affirming the consequent rather than a valid inference. https://aiindex.stanford.edu/report/
- Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?" *Proceedings of FAccT '21*, 610–623. A durable reference for the equivocation-on-"intelligence" and hasty-generalization patterns in AI discourse — arguing that fluent output can be mistaken for understanding. https://doi.org/10.

examples/ai-hype-discourse-fallacies-2024-2026.md

# Method in Action: Four Fallacies in the 2024–2026 AI Debate

> *Example for the [logical-fallacies](../SKILL.md) skill.*

The Linda problem shows the fallacy filter under laboratory conditions. This example runs the same **Fallacy Audit** on messy public discourse — the argument about artificial intelligence as it circulated across roughly 2024–2025. This is the environment the skill was built for: claims that *sound* authoritative, move fast, and carry high stakes, where naming the fallacy is only useful if you can also say what would actually settle the question.

We audit four representative arguments, each a real pattern that recurred across essays, interviews, and social threads in this period. No individual quote is reconstructed verbatim below; each argument is stated as a clean paraphrase of a widely-circulated *pattern* of reasoning, then run through the process.

---

## Step 1 — State each argument cleanly (premises → conclusion)

**Argument A (authority).**
- P1: The CEO of a leading AI lab says transformative AI is a few years away.
- P2: They run the lab and see the frontier models first.
- C: Therefore transformative AI is a few years away.

**Argument B (false dilemma).**
- P1: AI leads either to catastrophe ("doom") or to radical abundance ("utopia").
- P2: The doom scenario is implausible / the utopia scenario is implausible (whichever the speaker rejects).
- C: Therefore the other outcome is what we should expect.

**Argument C (hasty generalization).**
- P1: A single demo went viral showing a model doing task X impressively.
- C: Therefore models can now do X (and tasks like X) reliably.

**Argument D (affirming the consequent).**
- P1: If a system is intelligent, it will pass benchmark B.
- P2: This system passed benchmark B.
- C: Therefore this system is intelligent.

All four are genuine arguments (premises and a conclusion), so none collapses at Step 1 into "an assertion dressed as an argument." Good — that means the work is in the passes.

## Step 2 — Structural pass

**Argument D is a textbook formal fallacy: affirming the consequent.** The form is "If P then Q; Q; therefore P." That is invalid: Q can be true for reasons unrelated to P. Passing benchmark B is consistent with intelligence *and* with narrow pattern-matching, benchmark contamination (test items leaking into training data), or overfitting to the benchmark's format. The inference fails because P1 only licenses the reverse direction (intelligent → passes B), not (passes B → intelligent). Benchmark saturation across this period — models scoring very high on tests that older models failed — is exactly the observation that makes the invalid direction tempting.

**Argument C is hasty generalization.** One vivid, curated, possibly cherry-picked instance is generalized to reliable capability across a class of tasks. A viral demo is a maximally-filtered sample: the impressive run is the one that got posted. The inference from "did X once, on camera" to "does X reli

examples/tversky-kahnemans-linda-problem-1983.md

# Method in Action: Tversky & Kahneman's Linda Problem (1983)

> *Example for the [logical-fallacies](../SKILL.md) skill.*

To see logical-fallacy detection at full empirical force — not as debate-club gotcha but as an X-ray of how competent minds reason — the cleanest case in the modern literature is **Amos Tversky and Daniel Kahneman's 1983 paper** *"Extensional versus Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment"* (*Psychological Review*, Vol. 90, No. 4, pp. 293–315).

The experiment is famous, but its construction matters. Subjects were given a personality vignette:

> "Linda is 31 years old, single, outspoken and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations."

— Tversky & Kahneman, *Psychological Review* 90(4), p. 297. https://doi.org/10.1037/0033-295X.90.4.293

Subjects then ranked the probability of eight statements about Linda. Two of the statements were:

- (T) "Linda is a bank teller."
- (T∧F) "Linda is a bank teller and is active in the feminist movement."

The mathematical fact is **unambiguous**: P(T∧F) ≤ P(T), always, by the basic axioms of probability. The set of "bank tellers who are feminists" is a strict subset of the set of "bank tellers." This is not interpretation. It is definition.

The empirical result was equally unambiguous:

> "85% of subjects ranked T∧F as more probable than T... The conjunction effect was observed even among graduate students who had completed several courses in probability and statistics."

— Tversky & Kahneman, *Psychological Review* 90(4), p. 299.

That second sentence is the dagger. The error was not the property of the untrained; it was committed by **statistically-educated graduate students at Stanford and the University of British Columbia** — people who, if asked the abstract question "can a conjunction be more probable than its constituents?", would answer "of course not." Yet the *concrete* form of the question, dressed in a vivid representativeness-triggering personality sketch, broke their judgment systematically.

Tversky and Kahneman state the diagnosis precisely:

> "A conjunction cannot be more probable than one of its constituents. This fundamental rule of probability is violated... because the conjunction (a feminist bank teller) is more representative of Linda's personality than is the more inclusive category (a bank teller)."

— Tversky & Kahneman, *Psychological Review* 90(4), p. 297.

Here is what the case demonstrates for logical-fallacy detection:

**First**, the fallacy is **mechanical**, not malicious. No one in the experiment was trying to commit a fallacy. They were doing exactly what their minds spontaneously do — substituting the easy question ("does this story fit my stereotype of Linda?") for the hard question ("which set is mathematically larger?"). Aristotle's Sophists had to *work* to deploy fallacies; Tversky

Github ReposUpdated 2d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.