The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external websites, creating a surface for indirect prompt injection.
Ingestion points: The firecrawl_scrape tool (SKILL.md) is used to retrieve data from external web pages.
Boundary markers: Absent. While the workflow (SKILL.md) instructs the agent to cite sources and separate observations, it does not define explicit delimiters or instructions to ignore potential commands embedded in the scraped data.
Capability inventory: The skill can perform web scraping, capture screenshots, and execute various website health and security checks (strale_ssl_check, strale_header_security_check, etc.) through the sandbase gateway tools.
Sanitization: The instructions do not specify any methods for sanitizing or escaping the content retrieved from external URLs.

website-monitor