Description-Behavior Mismatch
Medium
- Confidence
- 96% confidence
- Finding
- 该模板并非通用学习笔记框架，而是预置了完整的 HTTP/2 攻击链、绕过分析、受影响范围和利用演示内容，明显超出“生成学习笔记样式 HTML”的声明用途。把攻击主题和利用细节固化进模板，会让技能在默认情况下产出进攻性内容，增加被直接用于传播或包装漏洞利用知识的风险。
Security audit
Note Skill
Security checks across malware telemetry and agentic risk
Overview
This skill is mostly a visual note generator, but one bundled template hard-codes actionable exploit-style security content and external network dependencies that users should review before installing.
Install only if you are comfortable reviewing and editing the bundled templates first. Replace the hard-coded attack-chain and exploit-command examples with neutral placeholders or defensive-only content, narrow the trigger words, and consider bundling fonts/icons locally if you need offline or privacy-sensitive HTML output.
SkillSpector
By NVIDIA
Vulnerability Patterns
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- 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 (7)
Context-Inappropriate Capability
High
- Confidence
- 99% confidence
- Finding
- 模板直接展示可针对指定主机执行的利用命令，并以低门槛、高效果的方式呈现攻击操作，这属于可操作的攻击指引而非单纯知识总结。放在一个可复用的前端模板中，会降低复制、传播和美化攻击说明的门槛，可能促进未授权利用。
Context-Inappropriate Capability
Low
- Confidence
- 94% confidence
- Finding
- The template fetches Google Fonts and Lucide CSS from third-party CDNs, which creates an external dependency and causes client-side network requests when the generated HTML is opened. This can leak viewer metadata such as IP address and user agent, and it also introduces supply-chain and availability risk if the remote assets change or are compromised.
Description-Behavior Mismatch
Medium
- Confidence
- 91% confidence
- Finding
- The skill claims to generate single-file HTML, but the template depends on external stylesheets for fonts and icons, so the output is not actually self-contained. In security-sensitive or offline review contexts, this can break rendering, trigger unexpected outbound connections, and violate assumptions users may make about safe local viewing.
Vague Triggers
Medium
- Confidence
- 94% confidence
- Finding
- The trigger list includes very broad everyday terms such as “笔记” and “note”, which can cause the skill to activate in many unrelated conversations. In an agent environment, this creates an attack surface for accidental invocation, context hijacking, or unintended application of the skill to sensitive content the user did not mean to format as an HTML note.
Missing User Warnings
High
- Confidence
- 97% confidence
- Finding
- 文件在给出利用说明和命令时，没有任何面向用户的合法授权、实验环境限定或防御研究声明，导致内容被理解为可直接照做的现实攻击指导。缺乏安全边界提示会显著增加误用和滥用概率，尤其是在技能会自动生成可分享 HTML 的场景下。
Missing User Warnings
Medium
- Confidence
- 96% confidence
- Finding
- The layout library includes a concrete privilege-escalation shell command inside a reusable HTML code-block template without any warning, sanitization guidance, or clear indication that it is unsafe example content. Because this skill generates study-note HTML and may be reused with security or exploit-analysis content, embedding a ready-to-copy `exploit && sudo ... bash -p` sequence increases the chance that dangerous commands are propagated or executed by users.
VirusTotal
VirusTotal findings are pending for this skill version.
Static analysis
No suspicious patterns detected.