The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to fetch and process data from 17 external search engines based on user queries. This creates a surface for indirect prompt injection where malicious instructions embedded in third-party websites could influence the agent's behavior once fetched.
Ingestion points: The skill utilizes the web_fetch tool to retrieve content from various search engine URLs defined in SKILL.md, config.json, and references/international-search.md.
Boundary markers: There are no instructions or delimiters provided to the agent to treat the fetched search results as untrusted content or to ignore embedded instructions.
Capability inventory: The agent has the capability to perform network operations (web_fetch) across all documented search engines.
Sanitization: No sanitization or filtering logic is defined for the content returned from the search engines before it is processed by the agent context.
[DATA_EXPOSURE]: The documentation in references/international-search.md includes examples of Google Dorks used to locate sensitive information on the public internet, such as intext:password filetype:txt. While this is a standard OSINT technique, it guides the agent toward searching for exposed credentials on the web.

multi-search-engine