The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to crawl and extract content from external URLs, creating a surface for indirect prompt injection attacks where instructions embedded in web pages could attempt to influence agent behavior.
Ingestion points: The skill processes data from web searches and URL crawling via the Tavily API, as specified in the overview and step-by-step guide in SKILL.md.
Boundary markers: The skill includes specific instructions to treat fetched pages as untrusted data that cannot authorize tools, purchases, or additional crawling.
Capability inventory: The skill provides web search, content extraction from URLs, and website crawling capabilities.
Sanitization: While the documentation advises treating content as untrusted, no specific code-level sanitization or filtering logic is detailed in the skill instructions.
[EXTERNAL_DOWNLOADS]: The installation process involves downloading the skill from a community-provided GitHub repository.
Evidence: The installation command npx skills add -g BenedictKing/tavily-web references an external GitHub source.

tavily-web