The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized operations were detected. The skill consists entirely of instructional text and templates for prompt engineering.
[PROMPT_INJECTION]: The skill includes templates designed to process untrusted external data (placeholders for {prompt} and {response}). While this constitutes an indirect prompt injection surface, the skill lacks any tool-calling or execution capabilities, rendering the risk negligible. The templates also employ clear structural delimiters (Markdown headers) to distinguish instructions from data.
[EXTERNAL_DOWNLOADS]: The documentation references reputable academic and technical sources such as arXiv and researcher blogs. No software packages, scripts, or binaries are downloaded or executed.

advanced-evaluation