ClawHub

胡田 OPC导师 三会治理系统

Security checks across malware telemetry and agentic risk

Overview

This is mostly a business-governance advice skill, but it includes under-scoped guidance for using personal government relationships to obtain approvals, public resources, and official support.

Review carefully before installing. Use this only with explicit user opt-in for governance analysis, remove or rewrite the government-relations guidance to require lawful public channels and compliance review, and avoid entering personal, confidential, regulated, or trade-secret information unless a protected handling process is in place.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (5)

Context-Inappropriate Capability

High
Confidence
97% confidence
Finding
该技能明确鼓励通过识别关键决策者、投其所好、建立私人信任与“利益共同体”来获取审批、公共资源和政府背书，这属于关系导向的影响力运作，容易滑向不当利益输送、权钱交易或规避正常审批程序。放在“治理支持系统”语境下，这类内容会被包装成方法论，增加被组织化复制和滥用的风险。

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
文档实际能力边界包含通过关系策略获取市场准入、政策支持、土地、资质、补贴及危机背书，这明显超出一般“治理支持”或内部组织顾问的合理范围。能力描述越界会误导调用方在高风险政商互动中依赖该技能，从而产生合规、反腐败和声誉风险。

Vague Triggers

Medium
Confidence
87% confidence
Finding
The trigger conditions are extremely broad and cover generic categories like strategy, operations, risk, talent, product, innovation, and consulting, which could cause this skill to activate for a wide range of ordinary business prompts. In an agent environment, overbroad auto-invocation can lead to prompt hijacking of user workflows, unnecessary data exposure to the skill context, and unintended reliance on this skill's governance framework where a narrower or more appropriate skill should have been used.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The template explicitly asks for name, company, department, and optional contact details, but provides no warning to avoid unnecessary personal, confidential, or regulated data. In a governance/advisory context, users may also include sensitive business, legal, or compliance information, increasing the chance of over-collection and privacy exposure.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger examples in the usage section are broad natural-language requests such as asking for a major decision review or a quick risk assessment. In an agent system that routes by semantic similarity, these generic phrases can cause the skill to activate for ordinary business conversation outside the user's intent, leading to unintended governance-style analysis or inappropriate handling of sensitive decision content. The business-governance context increases the risk because the skill is designed to influence high-impact strategic, legal, financial, and HR decisions.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal