The Agent Skills Directory

[SAFE]: The skill is designed for administrative and quality assurance purposes within a skill development environment. Its primary function is to scan local directories for SKILL.md files and validate their contents against a set of documented standards.
[COMMAND_EXECUTION]: While the skill requests the Bash tool, its described usage is limited to file discovery (globbing), searching (grep), and organizing information into report tables. There is no evidence of the skill executing arbitrary or untrusted code from external sources.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads local skill definition files but does not access sensitive system directories (like .ssh or .aws) or perform any network operations to send data externally.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a parser of other skills. While it ingests data from other skill files, it does so to perform static analysis (checking field presence, string length, and link validity). It does not appear to interpolate untrusted content into instructions that could trigger an agent takeover.

skill-audit