The Agent Skills Directory

[SAFE]: The skill implements strong safety protocols for cloud management, specifically distinguishing between read-only operations and mutating operations (create, update, delete).
[SAFE]: Instructions explicitly forbid the use of the --quiet flag on destructive commands to ensure that built-in gcloud safety prompts are not suppressed without user intent.
[SAFE]: The skill promotes security best practices, such as the principle of least privilege and the use of keyless authentication (Workload Identity) over long-lived service account keys.
[SAFE]: No evidence of prompt injection, data exfiltration patterns, or obfuscated content was found across the skill files.
[SAFE]: While the skill ingests external data (logs and resource states), it includes a mandatory confirmation workflow for any actions taken based on that data, effectively mitigating risks associated with indirect prompt injection.

gcloud-cli