The Agent Skills Directory

[SAFE]: No prompt injection or malicious instructions were found in the skill text or metadata. The instructions follow a standard, logical flow for email management.
[SAFE]: The skill operates on local data, such as drafts, configuration files, and CRM logs. It does not perform any unauthorized external network operations or demonstrate data exfiltration patterns.
[SAFE]: Strong safety guardrails are implemented, including a mandatory user confirmation step via AskUserQuestion. This ensures that a human must review the email content and recipient before the email-skills tool is invoked.
[SAFE]: The use of system tools is restricted to Bash for benign operations such as date calculations and checking file existence (test), which is appropriate for the skill's purpose.

send-outreach