ClawHub
Back to skill

Security audit

Apophenia Guard

Security checks for vulnerabilities and agentic risk

Overview

This skill is a markdown-only debugging reasoning aid that gives bias-check prompts and does not request system access, run code, or handle private data.

Before installing, expect this skill to be opinionated about debugging discipline and to interrupt with prompts when reasoning looks biased. I found no evidence that it executes code, accesses local data, calls external services, persists settings, or changes your environment.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
```
DETECTION SIGNALS:
├── Investigating only failed requests, not successful ones
├── "What's different about the broken ones?" without asking
│   "What's the same about the working ones?"
├── No control group in your investigation
└── Assuming the failure case is representative
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
DETECTION SIGNALS:
├── "That log entry is probably unrelated"
├── "Ignore that — it's a known issue" (is it though?)
├── Evidence that doesn't fit is labeled as noise without verification
└── The explanation requires you to ignore some of the data
```
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Vague Triggers

Medium
Confidence
90% confidence
Finding
This markdown file includes a 'When to Invoke' section, so SQP-1 applies. Phrases like 'At the start of any debugging session,' 'When you feel frustrated,' and 'When the explanation is getting more complex' are broad and subjective, without clear boundaries or exclusion conditions, which makes unintended invocation likely.

Static analysis

No suspicious patterns detected.