ClawHub
Back to skill

Security audit

Azure Cloud Inspector Free

Security checks across malware telemetry and agentic risk

Overview

This Azure inspection skill is broadly aligned with cloud auditing, but it needs review because it mixes read-only promises with modify/delete capability text and persists Azure inventory data locally.

Install only if you are comfortable letting the agent run Azure CLI commands in your current Azure login context. Treat it as a review-required skill: restrict it to a Reader-scoped account where possible, require explicit confirmation before any write or delete command, and check or remove the /tmp snapshot and report files after use if they contain sensitive resource details.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

High
Confidence
96% confidence
Finding
The document tells users the inspection will not modify Azure resources, but later capability text states the skill supports create/query/modify/delete operations. That contradiction can cause operators to trust the skill in read-only contexts while the agent may still attempt destructive actions if prompted, increasing the risk of unauthorized or accidental resource changes.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The trigger/capability scope is described so broadly that it overlaps with common cloud-assistant requests, which can cause the skill to activate in situations beyond the user's intended inspection task. In an agent environment with exec enabled, over-broad matching increases the chance of unintended command execution and misuse of a powerful cloud context.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill writes Azure inventory snapshots to local files under /tmp without a clear user-facing disclosure that resource metadata will persist on disk. Even if secrets are not intentionally collected, resource names, groups, tags, and topology can be sensitive and may be exposed to other local users, retained longer than expected, or ingested by backup/logging systems.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
Automatic generation of Markdown inspection reports to local disk can persist subscription names, resource counts, VM state, disk state, and exposure indicators without clearly notifying the user. This creates a confidentiality risk because cloud-environment details may remain on shared systems or developer workstations beyond the inspection session.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.