ClawHub
Back to skill

Security audit

SKILL1

Security checks across malware telemetry and agentic risk

Overview

This local file-cleanup skill mostly matches its purpose, but it needs review because it claims not to read file contents while duplicate detection does read files to hash them.

Review this before installing if you plan to run it on sensitive directories. Use the default preview mode first, supply only the intended target folder, and understand that clean/duplicates reads file contents locally to compute hashes even though it does not upload them.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The README makes a materially inaccurate security/privacy claim: duplicate detection by SHA-256 requires opening and reading file contents, so users may rely on a false assurance when processing sensitive local data. In a local file-management tool, misleading statements about what data is accessed can cause users to expose confidential files to processing they would otherwise avoid.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.