ClawHub
Back to skill

Security audit

huawei-cloud-bms-list

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed, read-only Huawei Cloud BMS inventory helper, with some operational cautions around broad trigger wording and a test script implementation.

Before installing, confirm you are comfortable granting the agent read-only visibility into Huawei Cloud BMS names, IDs, and statuses. Prefer a least-privilege IAM policy, do not hardcode AK/SK credentials, and avoid running the included test script with untrusted environment variables or edited test-vars.json values.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger wording around generic bare metal server listing can cause the skill to activate for broader BMS requests that exceed its stated read-only list scope. That creates a routing/intent-confusion risk where users asking for unsupported detail, admin, or lifecycle operations may be sent into the wrong skill path, leading to misleading outputs or accidental exposure of inventory data when the user intent was not properly constrained.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The ambiguous trigger phrase near the unsupported single-instance detail case may cause the skill to activate on requests for detailed server information rather than simple listing. In this context, the mismatch is more dangerous because the skill queries real tenant inventory, so incorrect activation can expose names, IDs, and statuses even when the user's request should have been refused or handled by a different, more narrowly scoped capability.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The script executes dynamically constructed shell commands with `eval`, and the command string includes variables such as `REGION` that are derived from environment variables or JSON input. An attacker who can influence those values can inject shell metacharacters and cause arbitrary command execution when the test script runs. In this skill context, the script is a test harness, but it may still be run in CI or developer environments with cloud credentials, making command injection materially dangerous.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.