The Agent Skills Directory

[SAFE]: The skill's behavior matches its stated purpose of humanizing text. It provides comprehensive guidelines for editing without introducing any malicious commands or hidden logic.
[PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection as it processes untrusted text from users or files. However, because the skill's capabilities are restricted to local file manipulation (Read, Write, Edit) and do not include network or system-level command execution, the risk is negligible.
[DATA_EXPOSURE]: The skill uses the Read tool to access user-specified writing samples for style matching. This is a legitimate use of the tool within the context of the skill's Voice Calibration feature and does not target sensitive system credentials or configurations.

humanizer