Back to skill

Security audit

Product Monster PRD

Security checks for vulnerabilities and agentic risk

Overview

This PRD-writing skill is mostly coherent, but it includes under-scoped local file generation, overwrite behavior, and remote diagram JavaScript that users should review before installing.

Install only if you are comfortable with a Chinese-focused PRD workflow that may create and overwrite diagram files near your PRDs. Do not run the documented Python command unless you have inspected and trust that local script, and prefer limiting generation to a specific PRD with backups. Treat generated HTML as network-active because it loads Mermaid from a CDN.

Vulnerability Patterns
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
references/flow-template.html:7
Finding

Unpinned Third-Party JavaScript Executed Without Integrity Verification

Content
View full analysis
``` ### Technical Analysis The flowchart template downloads and executes Mermaid JavaScript from the jsDelivr CDN whenever the HTML file is opened. The dependency is pinned only to the mutable major-version selector `@10`, rather than an exact immutable release. The script element also lacks a Subresource Integrity (`integrity`) hash. Consequently, the executable content delivered to users can change after the skill package has been reviewed. If the upstream package, publishing account, CDN, or dependency-delivery path is compromised, malicious JavaScript could be supplied and executed in the browser context of the generated flowchart. This is classified as `T08: Insecure Dependencies` because the risk arises from executing a remotely supplied third-party component without an immutable version or cryptographic integrity verification. The audited source does not demonstrate that the current Mermaid asset is malicious. ### Attack Path 1. A user creates or opens an HTML flowchart based on `references/flow-template.html`. 2. The browser requests `https://cdn.jsdelivr.net/npm/mermaid@10/dist/mermaid.min.js`. 3. An attacker compromises the relevant upstream package release, package-publishing account, CDN asset, or delivery infrastructure. 4. Because the version selector is mutable and no integrity hash is enforced, the browser accepts the altered response. 5. The malicious response executes as JavaScript in the flowchart document's browser context. Successful exploitation depends on compromise or malicious alteration of the external dependency-delivery chain. ### Impact Assessment An attacker controlling the delivered script could execute arbitrary JavaScript with the pri ...[truncated 891 chars]
Remediation
View remediation
``` 2. If CDN delivery is required, pin an exact immutable version and enforce Subresource Integrity: ```html ``` 3. Obtain the integrity hash from a trusted source or calculate and independently verify it against the reviewed release artifact. 4. Add a restrictive Content Security Policy that limits script execution and outbound connections to explicitly approved sources. 5. Establish a dependency-review process covering version updates, checksums, provenance, and vulnerability advisories. 6. Prefer an offline-capable template so generated diagrams do not depend on external runtime resources. ]]>
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (10)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest description states the skill must use '纯中文表达' and presents this as a mandatory requirement. This is a language-policy constraint applied by default rather than offered as a user choice, which can violate locale or language preference expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The rule says output '必须使用中文' except for a few standard terms, making Chinese the only allowed language regardless of user preference. Because the file does not offer an opt-in or alternative locale choice, this is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Manifest 将技能定位为 PRD 写作与评审工具,但文档进一步要求生成 PlantUML 图片、独立 HTML 文件,并保存到与 PRD 相同目录中。这已从内容生成扩展为文件产物管理与附属资源生成,和描述中的纯 PRD 撰写/评审语义不完全一致。

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

文档不仅要求写 PRD,还要求产出可直接在浏览器打开的独立 HTML 流程图文件,并规定样式表现。这属于文档之外的前端制品生成能力,对“撰写、扩写或评审 PRD”的核心目的并非直接必要。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

该技能名义上是 PRD 撰写/评审工具,但说明文档引入了自动扫描文档、编译 PlantUML、创建目录、生成并覆盖图片等本地文件处理能力,已经超出纯文本写作范畴。能力边界扩大后会引入额外攻击面:未经充分披露的文件写入、批处理和外部工具调用,都可能被用于修改工作区内容或触发不受期待的本地行为。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The description says the skill '强制使用' pure Chinese expression, which imposes a fixed language policy on users. This is a natural-language policy concern because it does not offer language choice or explain a justified regional limitation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

文档描述会自动创建 plantuml-images/ 目录、生成 PNG,并且自动覆盖已有图片,但没有在同一位置清晰告知存在文件修改和覆盖风险。对用户而言,这会造成“仅生成文档”与“实际修改本地文件系统”之间的认知偏差,增加误覆盖现有资源或污染仓库的风险。

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

文档明确引导用户进入固定本地路径并执行外部 Python 脚本,而且该脚本会扫描所有 PRD 文件并批量生成/重写输出。这种命令执行与目录扫描能力和 PRD 写作目的不匹配,一旦脚本被篡改、路径不适配或扫描范围过大,可能导致大范围文件修改,甚至成为执行本地恶意代码的入口。

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

手动重生成说明仅给出执行命令,但未充分提示该脚本会扫描所有 PRD 文件并重写图片输出,导致用户难以评估执行后果。批量扫描与重写在本地环境中尤其敏感,因为它可能影响多个项目文件、触发大量变更,且不易回滚。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The document declares lang="zh-CN" and uses Chinese-only title text, which imposes a specific language/locale in the skill artifact. The file does not indicate that this locale is optional, user-selectable, or required for a region-specific purpose, so it conflicts with the policy against forcing a language without opt-in.

Content

No source excerpt is available for this finding.