ClawHub
Back to skill

Security audit

代码质量检查专业版

Security checks across malware telemetry and agentic risk

Overview

This code-audit skill is mostly purpose-aligned, but it asks for powerful read, write, edit, command, and integration capabilities with broad activation language and limited user-control guidance.

Install only if you want an agent to run code-audit commands over your project and generate or edit files. Use it on trusted repositories, review commands before execution, keep external scanner settings disabled unless needed, and avoid broad automatic activation for normal coding tasks.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The skill claims that user input is not concatenated into commands, but its example workflow passes a user-controlled project path into command-driven processing. Even though the sample uses quoting, this still encourages execution over attacker-selected filesystem targets and creates a misleading security guarantee that may cause downstream unsafe implementations.

Vague Triggers

High
Confidence
91% confidence
Finding
The trigger condition is so broad that the skill could activate for routine development, debugging, or deployment tasks unrelated to secure auditing. In a skill with Read/Write/Edit/Bash capabilities, overbroad triggering raises the chance of unnecessary command execution, file modification, or collection of sensitive project data without clear user intent.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill advertises file writing, command execution, and API integration, but does not provide prominent user-facing warnings about side effects such as filesystem changes, report generation, network calls, or disclosure of code metadata to third parties. In a powerful agent context, missing warnings can lead to unintended system impact and unsafe operator trust.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.