The Agent Skills Directory

[SAFE]: The skill implements a standard OAuth device authorization flow for connecting to Agent Relay Cloud. It includes specific security best practices for handling sensitive tokens:
Recommends storing credentials in memory or private files with restricted permissions (directory 0700, file 0600) outside of repositories.
Prohibits printing, echoing, or dumping raw tokens or full authentication responses to logs, chat, or URLs.
Instructs the agent to validate api_url responses to ensure they do not point to unauthorized origins.
Provides guidance to prevent shell injection by advising against string interpolation when using system URL openers like open on macOS.
Uses official endpoints associated with the service and the author's own infrastructure (agentrelay.com).

signing-in-to-agent-relay-cloud