The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data retrieved from Google Search, Google News, and Google Maps results.
Ingestion points: Content is ingested via tools such as google_com__get_complete_s, news_google_com__get_rss_search, and google_com__read_page which fetch data from external domains.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore potentially malicious instructions embedded in the fetched web content.
Capability inventory: The skill possesses network-interaction capabilities; the potential for exploitation exists if the agent has access to other sensitive tools like file system writers.
Sanitization: There are no explicit sanitization routines mentioned to filter or escape the content returned from the external queries.
[EXTERNAL_DOWNLOADS]: The skill utilizes a remote Model Context Protocol (MCP) server located at https://unbrowse.ai/mcp/google.com. This is the vendor's official infrastructure for providing the skill's search and replay functionality.

google