ClawHub
Back to skill

Security audit

ROS 2 专家助手

Security checks across malware telemetry and agentic risk

Overview

This ROS 2 helper skill is coherent documentation, but it gives users unsafe system-install commands without enough safeguards.

Install only if you are comfortable reviewing commands before running them. Avoid running the one-line HTTP installer directly; prefer a verified HTTPS source, inspect the script first, and use a VM or container when possible. Back up APT configuration before changing sources, avoid deleting package-manager lock files unless you have confirmed no package process is running, and do not send unredacted logs, maps, voice data, locations, or API keys to external AI services.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The document instructs users to modify package sources, delete package manager lock files, and persistently alter shell startup files without warning about system stability, package database corruption, or unintended environment changes. In an installation guide, these actions are not inherently malicious, but presenting them without safeguards increases the chance that users will run risky commands blindly on production or shared systems.

Missing User Warnings

High
Confidence
98% confidence
Finding
The guide tells users to download and execute a remote script over plain HTTP using shell sourcing, which allows man-in-the-middle tampering and executes whatever content is served with the user's privileges. Because the command is framed as the primary install path and no integrity verification, HTTPS, signature check, or review step is provided, this is a genuine remote code execution risk.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The task encourages integrating voice recognition and sending error logs or related data to AI tools, but it does not warn that audio, transcripts, or logs may contain sensitive personal, environmental, or operational information. In a robotics setting, voice commands and logs can expose private speech, device state, filesystem paths, network details, or other sensitive telemetry if shared with third-party services without notice or minimization.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The document instructs users to obtain an LLM API key and send natural-language commands to an external LLM service, but it omits basic guidance on secret handling and external data exposure. In ROS 2/navigation workflows, prompts may include location data, map semantics, robot state, or user intent, and poor API key handling can also lead to credential leakage or unauthorized usage.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.