ClawHub
Back to skill

Security audit

Uncle Matt

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed security helper for routing approved API actions through a local broker, with no packaged executable code; the main caveat is its optional rude voice-pack style.

Install only if you want a restrictive local broker workflow. Before running the separate repo installer, review the broker and installer scripts, keep the broker bound to localhost, use Docker secrets or another protected secret store, and leave the voice packs disabled in professional or user-facing environments unless that tone is explicitly acceptable.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Natural-Language Policy Violations

Medium
Confidence
84% confidence
Finding
The skill explicitly allows the operator to enable profanity-heavy voice packs and to instruct the agent to prepend those lines during refusals or warnings, even when the end user did not opt in. This can create unwanted or abusive output toward users, weaken trust boundaries, and cause policy or UX violations, though it does not by itself create direct code-execution or secret-exfiltration risk.

Natural-Language Policy Violations

Medium
Confidence
94% confidence
Finding
The voice pack contains profanity and an aggressively forced persona style (for example, lines such as 'some stupid shit' and 'exactly how people get pwned') that can be emitted in refusals or warnings when enabled. This is not a code-execution or data-exfiltration flaw, but it is a real policy/compliance risk because it can cause the agent to generate hostile, unprofessional, or organization-inappropriate responses in security-sensitive interactions.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.