The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's HTML template (templates/page.html) references Google Fonts via fonts.googleapis.com and fonts.gstatic.com. These are well-known, trusted services used solely for typography and do not represent a security risk.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest content from Markdown design sources (specifications, architecture briefs) and render them into an HTML page. This creates a potential surface for indirect injection if sources are malicious.
Ingestion points: Markdown design sources read by the agent and transcribed into the PAGE JavaScript object within the generated HTML file.
Boundary markers: No specific boundary markers or "ignore" instructions are used; the agent is expected to interpret the Markdown content as data for the template.
Capability inventory: The skill allows the agent to write .html files to the local file system. The resulting HTML file contains interactive JavaScript for rendering diagrams and handling navigation.
Sanitization: The JavaScript machinery in templates/page.html uses innerHTML to render content from the PAGE object. While the documentation in references/template.md notes that fields are intended to support HTML, this means scripts could theoretically be injected if source documents are untrusted. However, within the context of a design review tool for humans, this is a functional characteristic rather than a malicious pattern.

walkthrough