The Agent Skills Directory

[SAFE]: The skill is composed of markdown documentation and evaluation scenarios. No executable scripts, suspicious commands, or malicious behaviors were identified.
[DATA_EXPOSURE]: The agent is instructed to read local context files, such as '.agents/product-marketing.md', to tailor email content. This is a standard practice for context-aware assistance and does not represent an exfiltration risk.
[EXTERNAL_DOWNLOADS]: The skill mentions several third-party email platforms and links to internal integration documentation. No unauthorized external downloads or network exfiltration patterns were found.

emails