The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interfaces with the 'gws' (Google Workspace) command-line tool to perform Gmail operations. This is a standard and expected behavior for a productivity skill authored by the service provider.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data for the email body, subject, and attachments. While this creates a potential surface for processing untrusted content, the skill explicitly includes a caution label for users to confirm actions before the command is executed, which follows security best practices for agent tools with write capabilities.
[SAFE]: All external references and requirements (such as the 'gws' binary) are consistent with the vendor's own ecosystem ('googleworkspace'). No malicious patterns, obfuscation, or unauthorized data access attempts were found.

gws-gmail-send