ClawHub
Back to skill

Security audit

Content Humanizer

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed writing assistant for auditing and rewriting AI-sounding drafts, with no evidence of hidden execution, data exfiltration, persistence, or destructive behavior.

Install only if you are comfortable with the skill using supplied drafts and an optional local marketing-context.md file for brand voice. Review outputs for language fidelity, especially because the fallback section says to default to Chinese if language detection fails.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description says the skill transforms robotic AI-generated content into genuinely human-sounding text. However, the supplied code is purely a scorer/auditor: it reads text, computes heuristic subscores, assigns a label like 'Sounds human' or 'AI fingerprint,' and prints recommendations. There is no text generation, editing, rewriting, or transformation capability implemented. This is a material primary-purpose mismatch, not just an implementation detail.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

Defaulting to Chinese during fallback without user consent can cause the model to transform or present content in an unintended language, creating integrity and safety risks for business, legal, or customer-facing text. In this skill's context, language fidelity is core to the task, so silent language switching can corrupt outputs, mislead reviewers, or leak process assumptions into production content.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill instructs reading marketing-context.md, which is a file-read capability, but it does not declare any tool scope or allowed-tools boundaries. Undeclared capabilities weaken permission transparency and can lead to broader-than-expected file access if the host agent infers or permits reads implicitly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

This markdown file explicitly defines trigger behavior, but the section header 'Proactive Triggers' and the instruction 'Flag these without being asked' are broad and lack clear activation boundaries. The file lists conditions to notice, but does not clearly constrain when the skill should activate versus when it should merely respond after explicit user use, increasing ambiguity for unintended invocation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest limits the skill to transforming existing AI-generated content and explicitly says it is not for initial content creation. However, the documented output artifact promises a 'Full rewrite,' which goes beyond light transformation and can amount to creating new copy from the source material.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.