ClawHub
Back to skill

Security audit

监控器

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to be a ping monitor, but it asks for command execution and describes broader file, API, automation, scheduling, and workflow behavior without clear boundaries.

Review before installing. Use it only for explicit ping or ICMP reachability checks, and do not grant broad command execution or file access unless you are comfortable with the agent running local diagnostics and reading local context. The artifact does not show malicious code, but its documented scope is too broad for a narrow monitor.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
92% confidence
Finding
The skill is marketed as a narrow ICMP health-check monitor, but later claims capabilities such as file handling, API integration, and command execution. This scope mismatch can cause an agent or user to authorize the skill for a low-risk monitoring task while unintentionally granting a much broader execution surface, increasing the chance of misuse or unsafe invocation.

Intent-Code Divergence

Medium
Confidence
90% confidence
Finding
The stated purpose is ICMP health checking, but the description broadens into unrelated automation, workflow, and general-purpose operational use. This inconsistency is dangerous because it obscures the real trust boundary of the skill, making over-permissioning and unintended execution more likely in contexts where users expect only simple network checks.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The invocation guidance is overly broad and includes many generic productivity and automation scenarios unrelated to ping monitoring. Overbroad trigger language can cause the skill to be selected in inappropriate contexts, exposing read/exec tools where they are unnecessary and increasing the risk of unintended system actions.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The documentation advertises command execution and file-processing behavior without prominent warnings, constraints, or safety boundaries. In a skill that exposes exec/read capabilities, this can lead users or orchestrators to approve actions without understanding the possibility of system modification, data access, or command abuse, making the skill materially more dangerous than a simple monitor.

Static analysis

No suspicious patterns detected.