The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill functions by processing untrusted Markdown input to generate HTML output, which creates a potential surface for indirect prompt injection if malicious instructions are embedded in the source text.
Ingestion points: The parseMarkdown and renderHtml functions ingest external markdown strings to create or render a MarkdownDocument AST (SKILL.md, references/ast-and-options.md).
Boundary markers: The library defaults to allowHtml: false and explicitly documentation notes that "unsafe executable protocols" are stripped from link and image targets.
Capability inventory: The skill provides text transformation and parsing logic; it does not include tools for arbitrary shell execution or persistent filesystem modifications.
Sanitization: The tool provides built-in protocol sanitization and requires explicit user configuration to enable raw HTML rendering.
[EXTERNAL_DOWNLOADS]: The instructions include steps to install the @tanstack/markdown package via pnpm. This is a standard dependency associated with the vendor's provided tools.

render-markdown