Back to skill

Security audit

学习指南网页生成器

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent study-guide webpage generator, but users should know it strongly favors Chinese output and CDN-loaded web assets.

Install only if you are comfortable with a Chinese-oriented study-guide generator that may activate broadly for educational webpage tasks. For private or sensitive notes, prefer editing the skill to bundle CSS/icons locally or use pinned integrity-checked assets instead of live CDNs.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T03 · Remote Payload Retrieval and Execution

Warning
Location
examples/example-guide.html:7
Finding

Mutable Third-Party JavaScript Is Retrieved and Executed at Runtime

Content
View full analysis
``` The corresponding Skill instructions require generated pages to use this remote resource: ```markdown 4. **单文件交付**:所有的 HTML、Tailwind CSS(通过 CDN)、FontAwesome 图标(通过 CDN)和 Vanilla JavaScript 逻辑必须封装在**同一个 HTML 文件**中。零外部依赖(除了 CDN 链接)。 ``` ```markdown - **技术栈**:严格使用 Tailwind CSS (CDN `https://cdn.tailwindcss.com`) 进行样式编写。 ``` ### Technical Analysis The page loads executable JavaScript from `https://cdn.tailwindcss.com` each time it is opened. The URL does not identify an immutable dependency version, and the script element does not include Subresource Integrity verification. Consequently, the code executed by the browser can change after the Skill and generated HTML have been reviewed. A compromise of the CDN, its publishing pipeline, DNS or network trust infrastructure, or the upstream resource could cause attacker-controlled JavaScript to execute in the page's browser context. This behavior is also mandated by `SKILL.md`, so pages subsequently generated according to the Skill are expected to inherit the same exposure. ### Attack Path 1. An Agent follows `SKILL.md` and creates an HTML study guide containing the Tailwind CDN script. 2. A user opens the generated page while connected to the internet. 3. The browser requests JavaScript from `https://cdn.tailwindcss.com`. 4. The CDN resource or its delivery path has been compromised or modified after the local HTML was reviewed. 5. The browser executes the modified response as JavaScript in the generated page's origin context. 6. The injected code reads or modifies page content and may access data available to scripts under that origin. 7. The injected code may send accessible infor ...[truncated 901 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
examples/example-guide.html:8
Finding

Font Assets Are Loaded from a Third-Party CDN Without Integrity Verification

Content
View full analysis
``` The Skill explicitly requires Font Awesome to be loaded through a CDN: ```markdown 4. **单文件交付**:所有的 HTML、Tailwind CSS(通过 CDN)、FontAwesome 图标(通过 CDN)和 Vanilla JavaScript 逻辑必须封装在**同一个 HTML 文件**中。零外部依赖(除了 CDN 链接)。 ``` ```markdown - **技术栈**:严格使用 Tailwind CSS (CDN `https://cdn.tailwindcss.com`) 进行样式编写。使用 FontAwesome (`https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.0.0/css/all.min.css`) 作为图标库。 ``` ### Technical Analysis The Font Awesome URL identifies version `6.0.0`, but the stylesheet is still accepted without a Subresource Integrity hash. The browser therefore trusts whatever content the CDN returns for that URL. A compromised CDN or delivery path could replace the expected stylesheet with attacker-controlled CSS. CSS is less capable than JavaScript, but hostile styles can alter or conceal page content, create deceptive overlays, redirect asset requests to attacker-controlled servers, and undermine the integrity of the user interface. Referenced font or image resources can also generate external requests and expose ordinary request metadata. Because the dependency pattern is required by `SKILL.md`, future generated pages are expected to include the same unverified external dependency. ### Attack Path 1. An Agent generates a study guide according to `SKILL.md`. 2. The generated page references the Font Awesome stylesheet on cdnjs. 3. A user opens the page, causing the browser to retrieve the stylesheet. 4. An attacker compromises the CDN resource or its delivery path. 5. The browser accepts the modified stylesheet because no integrity hash is provided. 6. The malicious CSS changes the vi ...[truncated 932 chars]
Remediation
View remediation
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (5)

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill’s trigger condition is excessively broad and mandatory, causing it to activate for many generic requests involving notes, webpages, cards, or teaching materials. This can hijack normal routing, override user intent, and increase the chance that unrelated tasks are forced through this skill, which is especially risky because the language says it 'must' be used even when the user did not explicitly request a study guide.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 58)May include surrounding context.

md
如果你需要参考一个完整的排版和高分交互代码示例,可以读取本技能目录下的 `examples/example-guide.html` 文件。它展示了以上所有的交互组件是如何融合在一个完美的单文件 HTML 结构中的。设计时请特别注意研究其 `<style>` 标签中关于 `.fade-in` 和 `.expand-con

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The skill metadata and content are written to enforce Chinese-language behavior and presentation style without offering user choice or documenting why locale restriction is necessary. This can cause the system to respond in an unintended language, degrade usability, and create policy-routing conflicts when users expect output in another language.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document declares lang="zh-CN", and all user-facing content is presented exclusively in Chinese. Under the policy, forcing a specific language or locale without user opt-in is a natural-language policy violation unless the restriction is explicitly justified or the user is given a choice.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The page imports Tailwind CSS and Font Awesome from third-party CDNs, which creates unnecessary external trust and runtime network dependencies for a static study-guide file. If those resources are unavailable, tampered with, or replaced upstream, the rendered page can break or execute attacker-controlled content in the user's browser.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.