ClawHub
Back to skill

Security audit

Humanizer Tool Free

Security checks across malware telemetry and agentic risk

Overview

The artifact is a text-humanization skill, but it declares shell and write authority plus broad save/delete/export wording that is not clearly scoped to text rewriting.

Review before installing. Use it only for user-provided text, and require explicit approval before any file save, export, modification, deletion, shell command, or network diagnostic. There is no evidence of exfiltration or destructive code, but the declared authority is broader than the text-humanization purpose needs.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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 (4)

Intent-Code Divergence

Medium
Confidence
93% confidence
Finding
The skill is described as a text-humanization tool, but the documentation also claims generic create/query/modify/delete/export/save capabilities. This mismatch expands the implied authority of the skill beyond its stated purpose and can mislead an agent into performing unintended file or data operations, especially when paired with WRITE/EXEC capabilities.

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
The skill declares exec capability even though its documented function is purely rewriting text in Markdown. Unnecessary shell execution materially increases attack surface because a prompt or skill flow could invoke system commands, network diagnostics, or local file access unrelated to the user’s editing task.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
Advertising EXEC and WRITE in a Markdown skill without clear warnings or guardrails normalizes potentially system-impacting behavior for a task that should only manipulate user-provided text. In context, this is especially dangerous because the same file later references network troubleshooting via ping, showing how elevated tools could be used outside the core function.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The documentation describes save/export/modify operations without warning the user that data may be changed or persisted. This can lead to silent overwrites, unintended exports, or user confusion about whether actions are in-memory edits versus filesystem changes.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.