The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation includes commands for component registration and setup using npx assistant-ui@latest add, which is the vendor's standard CLI workflow.
[EXTERNAL_DOWNLOADS]: The documentation instructs users to install several third-party libraries from the NPM registry, such as mermaid, remark-math, rehype-katex, and react-shiki, to support specialized markdown rendering features.
[PROMPT_INJECTION]: This finding identifies the attack surface for indirect prompt injection via rendered content.
Ingestion points: The MermaidDiagram component in references/latex-mermaid.md and the MarkdownText component in references/markdown-text.md ingest model-generated text directly.
Boundary markers: The skill relies on standard assistant-ui message part delimitation logic.
Capability inventory: The MermaidDiagram example uses innerHTML to inject SVG content and bindFunctions to attach event listeners to rendered diagrams.
Sanitization: Content rendering relies on the security protocols of the underlying libraries (mermaid, react-markdown). The alternative StreamdownTextPrimitive included in the skill documentation provides explicit allowedTags and linkSafety configurations for further mitigation.

markdown