ClawHub
Back to skill

Security audit

urlgo

Security checks across malware telemetry and agentic risk

Overview

This skill does what it says, but it gives an agent broad browser-control powers without enough safety boundaries for sensitive or authenticated sites.

Install only if you are comfortable letting an agent control a dedicated browser session. Avoid using it on banking, admin, internal, or other sensitive authenticated sites unless you explicitly direct each action, and close the launched browser or clear the urlgo profile when finished.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill explicitly enables opening arbitrary webpages, extracting page text/HTML, taking screenshots, typing into pages, clicking elements, and executing arbitrary JavaScript via a persistent browser session, but it provides no guardrails, consent requirements, origin restrictions, or warnings about handling sensitive data. In an agent setting, these capabilities can expose credentials, session cookies, private page contents, or internal web resources, and the instruction to use in-page navigation helps preserve authenticated state, which increases the chance of unintended data access or exfiltration.

Unvalidated Output Injection

High
Category
Output Handling
Content
def run_cmd(cmd, capture=True):
    """执行命令"""
    if capture:
        return subprocess.run(cmd, capture_output=True, text=True, shell=isinstance(cmd, str))
    else:
        return subprocess.run(cmd, shell=isinstance(cmd, str))
Confidence
92% confidence
Finding
subprocess.run(cmd, capture_output

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.