ClawHub
Back to skill

Security audit

作业

Security checks for vulnerabilities and agentic risk

Overview

This skill is aimed at job-application automation, but it combines external application submission, personal profile/API-key handling, and broad command execution with unclear safeguards.

Review this carefully before installing. Only use it with explicit user approval for each job application submission, avoid giving it broad API keys or unrestricted personal data, and do not assume its command execution is sandboxed unless your agent platform independently enforces that.

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 (2)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The skill claims that command execution happens in a 'safe sandbox', but the manifest only exposes a generic exec capability and provides no concrete sandboxing, confinement, or permission-boundary details. This can mislead users or calling agents into trusting shell execution that may actually run with normal host privileges, increasing the risk of unsafe command execution, data access, or environment compromise.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The skill's activation language is overly broad, invoking generic efficiency, automation, batch processing, and workflow-optimization scenarios far beyond job applications. In an agent ecosystem, this can cause the skill to be selected in unrelated contexts, exposing powerful tools like exec/read/grep/glob and potentially enabling unintended automation or risky actions without domain-appropriate guardrails.

Static analysis

No suspicious patterns detected.