ClawHub
Back to skill

Security audit

云运维编排器

Security checks for vulnerabilities and agentic risk

Overview

This skill is a cloud infrastructure operations guide that is mostly coherent, but it asks for broad execution/write authority over Terraform, Ansible, and cloud credentials while relying on safety controls that are described but not enforceable in the artifact.

Review this carefully before installing. Use it only in repositories and cloud accounts where the agent is explicitly allowed to run Terraform, Ansible, and cloud CLI commands. Require human review for plan, apply, reconcile, and destroy actions, especially in production, because the safety gates are documentation claims rather than enforced code in this artifact.

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 (3)

Intent-Code Divergence

Medium
Confidence
87% confidence
Finding
The skill claims command execution is restricted and that user input is not concatenated, but the document elsewhere exposes broad exec capability and parameterized operations without showing enforceable validation or an allowlist. This creates a trust gap: an agent or operator may rely on the safety claim and invoke powerful infrastructure actions under false assumptions, increasing the chance of unsafe command execution or destructive changes.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The invocation guidance is broad and blends cloud ops, security detection, compliance audit, vulnerability scanning, and encryption use cases without precise boundaries. Overbroad triggers can cause the skill to activate for loosely related prompts and then exercise read/write/exec capabilities in high-impact cloud environments, increasing the risk of unintended or excessive actions.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The keyword list is excessively broad and overlaps with common cloud, deployment, and infrastructure terms, which can cause opportunistic invocation on many unrelated requests. In a skill with exec, write, and cloud-operation semantics, overly permissive triggering materially raises the chance of accidental activation and unsafe infrastructure changes.

Static analysis

No suspicious patterns detected.