The Agent Skills Directory

[SAFE]: The skill documents a legitimate workflow for email management. It includes explicit safety sections that instruct the agent never to send emails and to only archive or draft based on specific user requests.
[SAFE]: It provides detailed remediation and setup instructions for 'deny' and 'ask' rules in agent harnesses (like Claude Code) to prevent the accidental execution of sensitive tools such as send_message, reply, or trash_message.
[SAFE]: External references target well-known and trusted services, specifically the Google Gmail MCP server (gmailmcp.googleapis.com). No suspicious remote code execution or data exfiltration patterns were detected.

email