The Agent Skills Directory

[NO_CODE]: The skill consists entirely of Markdown instructions and a JSON configuration snapshot. No executable scripts (Python, JavaScript, etc.) are included in the skill package.
[COMMAND_EXECUTION]: The skill mentions using a Python script validate-audit-artifact.py and standard git commands (git rev-parse) for environment setup, but these are standard development workflow practices and do not involve executing untrusted or remote code.
[DATA_EXPOSURE]: The skill defines a specific persistence path (memory/audits/ad/) for its audit reports and requires explicit authorization before writing. It correctly distinguishes between platform-reported data and own-data truth sets, recommending standard reconciliation rather than sensitive data exfiltration.
[SAFE]: The skill refers to external dependencies within its own repository structure and follows a 'fail-closed' standalone policy when required root components are unavailable. All referenced URLs point to the author's own GitHub repository (github.com/aaron-he-zhu/*), which is consistent with the author context.

ad-account-auditor