The Agent Skills Directory

[SAFE]: Functional Safeguard. The skill explicitly instructs the agent to stop before authentication or command invocation, as the underlying training backend is currently unavailable. This prevents the agent from attempting to execute commands that are not yet functional.
[SAFE]: Security-First Design. The skill mandates strict future security controls, such as prohibiting the inclusion of secrets in manifests or logs, rejecting symlinks, and using cryptographic digests to ensure manifest immutability.

ito-training