The Agent Skills Directory

[SAFE]: The skill includes a comprehensive 'Security Boundaries' section that explicitly addresses potential attack vectors. It mandates that all user-provided values, such as namespaces and workload names, must be validated against strict regex patterns (^[a-z0-9]([a-z0-9-]{0,61}[a-z0-9])?$) before being used in any auto-discovery commands, effectively preventing command injection.
[SAFE]: The skill implements best practices for credential management. It forbids hardcoded secrets, prevents the agent from requesting sensitive information from the user, and instructs the use of Kubernetes Secrets via valueFrom.secretKeyRef. It also provides secure guidance on secret creation (using --from-file instead of --from-literal) to avoid exposing credentials in shell history.
[SAFE]: The skill maintains a clear separation between data discovery and cluster modification. While it uses read-only commands (kubectl get, kubectl config) to gather context, it strictly prohibits the agent from executing modifying actions like kubectl apply or helm install, ensuring that all generated YAML and cluster changes remain under manual human control.

mirrord-kafka