alibabacloud-network-diagnose

• Outcome: Review
• Latest audit: Jul 30, 2026 · 3:37 AM UTC
• Version: 0.0.1

Overview

The skill is a coherent Alibaba Cloud network diagnostic tool, but it can expose VPN pre-shared keys and automatically install CLI plugins, so it needs review before use.

Install only if you are comfortable letting the skill run Alibaba Cloud CLI read-only diagnostics with your configured account. Avoid using it for VPN Gateway cases until PSK output is removed or redacted, and prefer requiring explicit approval before any CLI plugin installation or multi-region probing.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (6)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 95% confidence
• Finding: The skill instructs the agent to execute shell commands, read environment state, and read/write files, but the metadata declares no permissions. This creates a capability/expectation mismatch that can bypass user or platform review, making it easier for the skill to access credentials, local state, and generated artifacts without explicit approval boundaries.

Context-Inappropriate Capability

Medium

• Confidence: 90% confidence
• Finding: The skill can modify the host by automatically installing Alibaba Cloud CLI plugins when error text suggests one is missing. For a diagnostic skill, host package/plugin installation expands capability and changes system state unexpectedly, which is risky in shared runners or tightly controlled environments.

Description-Behavior Mismatch

High

• Confidence: 99% confidence
• Finding: The function returns the VPN tunnel pre-shared key (`Psk`) as part of diagnostic output. In a diagnostic skill, this materially increases risk because the value may be printed to console, logged, stored in transcripts, or exposed to users/operators who only need connectivity state, turning a troubleshooting tool into a secret disclosure path.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: This routine collects full tunnel IKE configuration and includes the PSK in the returned structure without clear operational need for network diagnosis. Because the script later JSON-serializes results, the secret can propagate into terminal output, automation logs, support artifacts, or agent conversation history, causing credential exposure beyond the original API boundary.

Missing User Warnings

Low

• Confidence: 86% confidence
• Finding: The degradation path says to try several regions in sequence when the region is uncertain, which broadens the scope of cloud queries beyond the user's confirmed target. Even though the operations are described as read-only, this can enumerate assets across multiple regions and violate least-knowledge expectations or surprise users about account-wide probing.

Natural-Language Policy Violations

Medium

• Confidence: 79% confidence
• Finding: Mandating an exact Chinese sentence in the final report without user language selection can cause the agent to emit content the user may not understand and reduces transparency of a security diagnosis. In a troubleshooting skill this is mainly a safety/UX issue, but it can impair informed consent and accurate interpretation of important routing conclusions.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.