cognitive-fallacies-guard
Detects and prevents visual misleads, cognitive biases, and data integrity violations in visualizations, dashboards, reports, and presentations. Audits charts for honesty, diagnoses misinterpretation causes, and provides specific fixes. Invoke when user mentions chartjunk, misleading chart, truncated axis, data integrity, visual deception, 3D chart problems, cherry-picking data, or needs to audit visualizations for accuracy. For general design evaluation, use `design-evaluation-audit`. For cognitive foundations, use `cognitive-design`.
How do I install this agent skill?
npx skills add https://github.com/lyndonkl/claude --skill cognitive-fallacies-guardIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill is a set of instructional guidelines and audit checklists designed to identify cognitive fallacies and visual distortions in data presentations. It contains no executable code, remote dependencies, or suspicious network activities. The primary risk is a theoretical surface for indirect prompt injection through user-provided chart descriptions, but the skill lacks any capabilities to exploit such an injection.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
- ZeroLeakspass
Score: 93/100 · 2 sections analyzed
What does this agent skill do?
Cognitive Fallacies Guard
Table of Contents
Overview
Visualizations are persuasive — common mistakes cause systematic misinterpretation, not just aesthetic failures. This skill scans for visual misleads (chartjunk, truncated axes, 3D distortion), checks for cognitive bias exploitation (confirmation bias reinforcement, anchoring, framing manipulation), and verifies data integrity (honest axes, complete data, fair comparisons).
Related skills: design-evaluation-audit for general design evaluation, cognitive-design for cognitive foundations, d3-visualization for creating visualizations, visual-storytelling-design for data stories.
Fallacy Audit Workflow
Time: 15-30 minutes
Copy this checklist and track your progress:
Fallacy Audit Progress:
- [ ] Step 1: Scan for Visual Misleads
- [ ] Step 2: Check for Cognitive Biases
- [ ] Step 3: Verify Data Integrity
Step 1: Scan for Visual Misleads
Check for chartjunk, 3D effects, truncated axes, volume illusions, and inappropriate chart types. These are the most common and visible fallacies.
Resource: Fallacies Catalog — Sections 1-2 (Visual Noise, Perceptual Distortion)
Step 2: Check for Cognitive Biases
Look for confirmation bias reinforcement, anchoring effects, and framing manipulation. These are subtler but can significantly influence interpretation.
Resource: Fallacies Catalog — Section 3 (Cognitive Bias Exploitation)
Step 3: Verify Data Integrity
Confirm honest axes, complete data, fair comparisons, proper context, and no spurious correlations. This is the most critical layer.
Resource: Detection Patterns — Integrity Principles and Quick Scan Checklist
Path Selection Menu
Path 1: Visual Misleads Scan
Choose this when: Checking for chartjunk, 3D effects, truncated axes, and encoding problems.
→ Go to Fallacies Catalog — Sections 1-2
Path 2: Cognitive Bias Check
Choose this when: Looking for bias reinforcement in dashboard design, presentation framing, or data selection.
→ Go to Fallacies Catalog — Section 3
Path 3: Data Integrity Verification
Choose this when: Verifying completeness, honesty, and context of data presentation.
Quick Reference
5 Integrity Principles
- Honest Axes — Bar charts start at zero; uniform scale intervals; clear labels
- Fair Comparisons — Same scale for compared items; no dual-axis manipulation
- Complete Context — Full time period shown; baselines provided; denominators clarified
- Accurate Encoding — Visual proportional to numerical; no volume illusions; 2D design
- Transparency — Data sources cited; limitations acknowledged; methodology stated
Quick Severity Guide
- Severity: High — Integrity violations (truncated bars without disclosure, cherry-picked data, implied causation)
- HIGH: Perceptual distortions (3D effects, volume illusions, missing denominators)
- MEDIUM: Bias reinforcement (one-sided framing, anchoring order, confirmation bias layout)
- LOW: Visual noise (excessive gridlines, decorative elements, ornamental borders)
Guardrails
Scope boundaries: This skill detects visual misleads, identifies cognitive bias exploitation, verifies data integrity, and provides specific fixes for each fallacy found. It does not create designs, evaluate general usability, teach cognitive theory, or assess aesthetic quality.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/lyndonkl/claude/cognitive-fallacies-guard">View cognitive-fallacies-guard on skillZs</a>