Security audit

Azure管理专业版

Security checks across malware telemetry and agentic risk

Overview

This Azure administration skill is not plainly malicious, but it requests broad cloud-changing authority and credential handling without enough scoping or safeguards.

Review this skill before installing. Use it only in Azure subscriptions you control, prefer least-privilege or managed identity credentials, do not store client secrets in local project files, and require an explicit plan or dry run before any deploy or optimization command changes resources or costs.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Medium

Confidence: 95% confidence
Finding: The skill says API keys should not be hardcoded, but also instructs that credentials are stored in a local configuration file. That contradiction can normalize insecure secret handling and lead operators to place long-lived Azure credentials in plaintext files, which are commonly leaked through source control, backups, or host compromise.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The skill description uses broad trigger language such as security detection, compliance audit, quality check, and encryption protection without tight activation boundaries. In an agent environment, this can cause the skill to be invoked for loosely related requests and expose powerful read/write/exec behavior in contexts where the user did not clearly intend Azure administration actions.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The capability scope lists an extremely broad set of keywords, including generic terms like 'Use', 'when', 'quality check', 'automation', and 'workflow'. Such expansive matching increases accidental activation risk and can cause an agent to route unrelated requests into a skill that supports command execution and cloud-changing operations.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The deployment examples generate templates, write files, and deploy infrastructure but do not prominently warn about creating or modifying cloud resources or overwriting local output paths. In this context, the skill advertises exec and write tools, so users or agents may trigger costly or destructive changes without understanding the side effects.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The optimization example includes an apply-style flag for recommendations without a clear safety warning about changes to live environments. Even with '--dry-run' shown nearby, presenting apply semantics in cost optimization can mislead agents or operators into approving resource modifications that affect availability, performance, or billing.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.