The Agent Skills Directory

[SAFE]: The skill's instructions are purely focused on linguistic editing and do not contain any malicious patterns such as prompt injection, data exfiltration, or persistence mechanisms. The capabilities requested (Read, Write, Edit, Grep, Glob) are appropriate for its stated purpose as a writing tool.
[EXTERNAL_DOWNLOADS]: The skill contains a reference to an official Wikipedia page regarding AI writing. As Wikipedia is an established and well-known service, this external reference is considered safe.
[SAFE]: While the skill processes user-provided text (ingestion point) and has file modification tools, it does so within the context of a stylistic editor. No evidence of unsafe data interpolation or sanitization bypasses was found, and the behavior is consistent with the primary skill purpose.

humaniser