The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a pattern that processes markdown files from an external source and renders them in the UI, creating an attack surface for indirect injection if the storage is compromised.
Ingestion points: The BlogStorageService.cs class in the API project retrieves markdown file content from an Azure File Share.
Boundary markers: The system uses triple-dash --- delimiters to distinguish between YAML metadata and the markdown body content.
Capability inventory: The skill has the capability to render raw HTML to the user's browser via the Blazor MarkupString type.
Sanitization: While it uses the Markdig library for processing, there is no evidence of an explicit HTML sanitization step (e.g., using a library like HtmlSanitizer) before the content is rendered in the Post.razor component.
[DYNAMIC_EXECUTION]: The skill generates and injects HTML content into the application DOM at runtime.
Evidence: In frontend-components.md, the Post.razor component uses renderedContent = Markdown.ToHtml(post.Content, pipeline); followed by @((MarkupString)renderedContent). This bypasses Blazor's default HTML encoding to render the markdown output, which is a form of dynamic content generation that requires strict input validation.

blazor-blog-feature