Security audit
Security checks for vulnerabilities and agentic risk
This skill is a PDF-formatting guide with no executable payload, persistence, credential handling, or hidden data movement.
Before installing, treat it as a specialized PDF authoring workflow: it will steer agents toward a specific html_to_pdf pipeline and print-oriented CSS conventions. The main caveat is minor ambiguity in its CSS-counter guidance, not a security risk.
The documentation says not to rely on CSS counters and explicitly lists counter-reset, counter-increment, and counter() as disallowed. However, later examples use counter(page) and counter(footnote) for page headers and footnote markers, which contradicts the earlier blanket guidance and could mislead an agent implementing the skill.
The language policy mandates 'Chinese user query -> Chinese content' and 'English user query -> English content' rather than offering a language choice. This is a natural-language locale policy constraint that could override user preference unless they explicitly restate it.
The file warns that Paged.js should not be injected manually because the runtime pipeline handles loading. Later sections rely on paged-media features such as @page, target-counter, and float: footnote, which are compatible with the stated pipeline but leave the earlier instruction potentially confusing about when Paged.js-backed features are expected to work.
No suspicious patterns detected.
