ClawHub
Back to skill

Security audit

域名DNS管理专业版

Security checks across malware telemetry and agentic risk

Overview

This skill is for legitimate DNS operations, but it can perform high-impact domain changes with broad invocation wording and insufficient explicit confirmation safeguards.

Review this skill carefully before installing. Use it only for explicit DNS/domain tasks, with least-privilege API tokens, test domains first, and require a preview plus explicit confirmation before any record, nameserver, redirect, DNSSEC, Worker, rollback, cron, or bulk operation.

SkillSpector

By NVIDIA
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 (3)

Intent-Code Divergence

Medium
Confidence
84% confidence
Finding
The trigger conditions are broader than the stated DNS purpose, covering generic monitoring, log analysis, alerting, and deployment management. In an agent ecosystem, overly broad invocation criteria can cause the skill to be selected in unrelated operational contexts, increasing the chance of unnecessary command execution or unintended infrastructure changes.

Vague Triggers

High
Confidence
91% confidence
Finding
Overly broad and mismatched trigger conditions raise the risk that an autonomous agent invokes this skill for general ops requests outside its safe scope. Because the skill exposes exec/write capabilities and supports destructive DNS changes, unintended invocation could lead to service disruption, misconfiguration, or unauthorized operational actions.

Missing User Warnings

High
Confidence
94% confidence
Finding
The skill describes automatic execution of high-impact DNS operations such as changing A records, switching nameservers, bulk redirects, and rollback actions without a prominent confirmation step. In a tool-enabled agent context, this can translate natural-language requests directly into destructive infrastructure changes, creating a significant risk of outages, traffic hijacking, or accidental domain takeover scenarios.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.