ClawHub
Back to skill

Security audit

vmware-policy

Security checks across malware telemetry and agentic risk

Overview

This looks like a legitimate VMware policy and audit helper, but its broad fail-open and bypass modes deserve review before relying on it for enforcement.

Review this before installing if you expect it to enforce controls by default. Create and test ~/.vmware/rules.yaml, install PyYAML where required, restrict who can set VMWARE_POLICY_DISABLED, and treat the audit database as potentially sensitive because it records tool parameters, results, user, agent, and operation metadata.

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 (3)

Vague Triggers

Medium
Confidence
86% confidence
Finding
The skill declares that it auto-activates whenever any VMware skill tool is called or when broad audit/policy-related user intents are detected. Because this package is a shared dependency that wraps 156+ tools, such broad activation increases the chance that policy/audit logic runs in unintended contexts, expanding the trusted computing base and making prompt- or context-triggered behavior harder to reason about. The risk is amplified because it sits in the execution path of many other skills.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The CLI reference explicitly documents `VMWARE_POLICY_DISABLED=1` as a way to bypass policy checks, but does not warn that this disables a core security control across the VMware skill family. In this skill context, policy enforcement is a central guardrail for 156+ tools, so normalizing a bypass mechanism in user-facing docs materially increases the chance of intentional misuse or accidental unsafe operation.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The guide documents two insecure fail-open states without adequately warning users: setting `VMWARE_POLICY_DISABLED=1` bypasses policy checks, and absence of `~/.vmware/rules.yaml` allows all operations. In a shared policy/audit library that wraps many VMware tools, this can lead operators to assume enforcement exists when it may be fully disabled, resulting in unauthorized or risky actions being permitted across the skill family.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.