The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: Provides instructions for installing the Helm CLI by downloading and executing a script from the project's official GitHub repository.
[DATA_EXFILTRATION]: Documents standard administrative procedures for retrieving and decoding sensitive configuration data stored within Kubernetes Secrets.
[COMMAND_EXECUTION]: Enables interactive shell access, resource creation, and administrative debugging within cluster workloads using kubectl exec, kubectl run, and kubectl debug.
[EXTERNAL_DOWNLOADS]: Supports the application of Kubernetes resource manifests directly from remote URLs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from the cluster environment.
Ingestion points: Cluster data retrieved via kubectl logs, kubectl get events, and resource descriptions.
Boundary markers: Not present; there are no instructions to treat log or event data as untrusted or to ignore embedded commands.
Capability inventory: Broad cluster administrative permissions through kubectl, helm, and kustomize including file system modification and network operations.
Sanitization: None; the agent processes the raw output of cluster logs and events directly into its context.

kubernetes-operations