The Agent Skills Directory

[SAFE]: The skill serves as a security control mechanism, providing a clear policy to prevent autonomous agents from performing unauthorized external actions when human oversight is absent.
[SAFE]: The instructions emphasize 'reversible-and-invisible' criteria for unattended work, specifically flagging high-risk commands such as --force, --no-verify, or --admin as actions to be avoided during unsupervised periods.
[SAFE]: All external resource references point to the vendor's official GitHub repositories (thrillmade/agent-skills) and well-known documentation sources, which is consistent with standard development practices.
[SAFE]: The skill implements strict verification steps, such as using git ls-remote to ensure remote repositories remain unchanged unless explicitly authorized, protecting against accidental or malicious modifications to shared codebases.

unattended-operation