ClawHub

胡田 OPC导师 政策工具包

Security checks across malware telemetry and agentic risk

Overview

The skill is a coherent policy-application helper, but it may handle sensitive identity, payroll, banking, credit, and business records without enough privacy or sharing limits.

Install only if you intend to use an agent for sensitive policy-application work. Require explicit approval before web searches, file creation, or handoff to other skills; redact unnecessary IDs, bank, payroll, credit, and employee details; and decide where generated Markdown and JSON outputs may be stored, shared, and deleted.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The activation text says users can activate the skill by directly describing their current application scenario, which is broad enough to match ordinary policy discussion rather than an explicit opt-in to a specialized workflow. In practice this can cause accidental invocation, making the agent collect business details, perform external lookups, or generate application artifacts when the user only wanted general advice.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill explicitly defines generation of reports, material checklists, JSON records, and handoff artifacts, but it does not warn that enterprise qualification data and personal information may be persisted in files or passed to downstream skills. Because the toolkit handles IDs, financial reports, social security records, employee rosters, and qualification records, silent artifact creation increases the risk of unnecessary retention, oversharing, and privacy or compliance violations.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal