The Agent Skills Directory

[SAFE]: The skill is composed entirely of instructional text guiding the agent on search strategies and information synthesis. It contains no scripts, commands, or hidden data.
[NO_CODE]: No Python, Node.js, or shell scripts are included in the skill content. There are no dependency requirements or installation steps.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from web search results. This creates an attack surface where malicious content on websites could attempt to influence the agent. The instructions mitigate this risk by requiring source citations, primary source preference, and result corroboration. Given this is the primary purpose of the skill, the risk is considered inherent to the operation and managed by the prompt's guidelines.

web-search