The Agent Skills Directory

[PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface as it is designed to retrieve and process content from the open web.
Ingestion points: scripts/search.mjs and scripts/extract.mjs ingest snippets and full-page text from arbitrary external URLs.
Boundary markers: Results are structured using standard markdown headers and lists, but the skill does not implement specific delimiters or instructions to the agent to disregard potential instructions embedded in the retrieved web content.
Capability inventory: The skill scripts lack dangerous capabilities; they do not perform file system writes, execute shell commands, or use dynamic execution functions like eval() or exec().
Sanitization: Basic string conversion and whitespace trimming are applied to the outputs, though no semantic sanitization of the fetched content is performed.
[DATA_EXFILTRATION]: The skill initiates network requests to api.tavily.com to provide its search and extraction features. This represents legitimate communication with a well-known service provider required for the skill's stated purpose.

tavily