ClawHub
Back to skill

Security audit

Narrative Fallacy

Security checks for vulnerabilities and agentic risk

Overview

This is a Markdown-only reasoning skill that guides narrative-fallacy analysis and does not request sensitive access, execution, or persistence.

Installing this skill should mainly add a structured thinking aid. Reviewers should be aware it contains links to related skills and public reference material, and users who do not want agents editing installed skill documents should treat the suggested addition of observed examples as optional/manual only.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · examples/talebs-911-example-business-history-critique.md (reported line 26)May include surrounding context.

md
**Post-mortems with neat root causes.** Engineering, medical, and management post-mortems regularly produce "the root cause was X" findings that, on closer examination, are narrative-fallacy products. The genuine causal structure of complex-system failures is usually a confluence of many small factors, each individually insufficient. The "root cause" framing strips out the others. Decisions based on "we fixed the root cause" then leave the system vulnerable to the other unfixed contributors.

The medical / aviation / nuclear-safety community has been moving toward "blameless postmortem" and "Swiss-cheese model" framings explicitly to resist the narrative-fallacy temptation. See [`premortem`](../skills/premortem/SKILL.md) and the broader work on systems safety (Reason 1990; Hollnagel 2014).

The intellectual lineage of the framework:

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · examples/talebs-911-example-business-history-critique.md (reported line 30)May include surrounding context.

md
The intellectual lineage of the framework:

**Kahneman's *Thinking, Fast and Slow* (2011)** chapter 19 ("The Illusion of Understanding") treats the same phenomenon under the term "illusion of understanding," explicitly linking it to hindsight bias and the illusion of validity. Kahneman attributes the cognitive root to System 1's automatic story-construction (see [`dual-system-thinking`](../skills/dual-system-thinking/SKILL.md)).

**Yuval Harari's *Sapiens* (2014)** ch. 2 makes the broader evolutionary-psychology case that humans are "the storytelling ape" — our cognitive architecture evolved to construct and remember narratives because narratives enable large-scale social coordination. The same evolutionary advantage produces the systematic error of mistaking constructed narratives for causal explanations.

Static analysis

No suspicious patterns detected.