ClawHub
Back to skill

Security audit

Pg Update

Security checks across malware telemetry and agentic risk

Overview

This ProxyGate update skill is mostly coherent, but it can change global packages and installed agent skills from broad or proactive triggers without clear confirmation.

Install only if you are comfortable letting the agent assist with ProxyGate updates that may change a global npm package, project dependencies, and installed ProxyGate skills. Before running it, explicitly confirm that the update is for ProxyGate, ask to see the exact commands and target versions, and do not treat generic prompts like `upgrade cli` as approval to update ProxyGate.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Confidence
78% confidence
Finding
The trigger conditions are broad enough to cause the skill to activate proactively at session start whenever an update notification appears, which can steer the agent into package-management actions without an explicit user request in that moment. In a security-sensitive environment, overly broad activation increases the chance of unreviewed system changes or supply-chain exposure from automatic update behavior.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The skill instructs deletion of a file in the user's cache directory without warning, confirmation, or explanation of side effects. Even though the target is narrow, normalizing silent deletion in a skill increases the risk of unintended data loss and conditions users to accept destructive filesystem operations without review.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.