The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides extensive capabilities to manage Kubernetes clusters through the kubectl CLI, including resource creation, modification, deletion, and interactive shell access to containers.\n
SKILL.md documents powerful commands such as kubectl exec, kubectl edit, kubectl apply, and kubectl port-forward.\n
Automated utility scripts like cluster-health-check.sh, namespace-cleanup.sh, and pod-debug.sh facilitate multi-step cluster operations.\n- [CREDENTIALS_UNSAFE]: The documentation and resource templates contain example credentials used for illustrative purposes.\n
SKILL.md defines a Kubernetes Secret template with generic values like user:pass and secret-key-value.\n
references/best-practices.md includes a configuration example using DATABASE_PASSWORD: "secret123".\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from the Kubernetes cluster environment, which could be influenced by external actors.\n
Ingestion points: The agent is instructed to read pod logs and cluster events using kubectl logs and kubectl get events in scripts/pod-debug.sh, scripts/cluster-health-check.sh, and the troubleshooting guide.\n
Boundary markers: There are no specific delimiters or instructions to treat the output of logs or events as untrusted data.\n
Capability inventory: The skill possesses broad administrative capabilities, including the ability to execute code in containers and modify cluster resources across scripts like namespace-cleanup.sh.\n
Sanitization: External data from logs and events is processed without sanitization or validation against unexpected instructions.

kubernetes-ops