ClawHub
Back to skill

Security audit

Security Audit (Sona)

Security checks across malware telemetry and agentic risk

Overview

This appears to be a defensive local security-audit skill, with one reliability concern about its shell wrapper reporting success even when the audit runner fails or finds issues.

Reasonable to install for local defensive auditing. Before using it as a release or CI gate, verify how failures are signaled and configure your workflow to fail when the JSON result indicates findings, malformed output, or runner errors.

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
97% confidence
Finding
The script captures the audit runner's exit status in RC but then unconditionally exits 0, so any caller relying on process status will treat a failed or finding-bearing audit as success. In a security-gating skill, this fail-open behavior can allow vulnerable or noncompliant skills/repos to proceed despite detected issues, especially if downstream integrations ignore or fail to parse the JSON `.ok` field.

VirusTotal

48/48 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.