The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of kubectl and flux CLI tools to perform operations across the Kubernetes cluster. Many commands use the -A or --all-namespaces flag, granting the agent high visibility into resource statuses and logs across all namespaces in the cluster.\n- [CREDENTIALS_UNSAFE]: A diagnostic command is provided to specifically decode and display registry credentials: kubectl get secret -n flux-system flux-system -o jsonpath='{.data.\.dockerconfigjson}' | base64 -d. This exposes sensitive Docker registry authentication data in the command output, which could be logged or seen by unauthorized parties.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads and processes untrusted data from external sources such as pod logs and Kubernetes event messages. \n
Ingestion points: Kubernetes pod logs and cluster events via flux logs and kubectl get events.\n
Boundary markers: None identified; output is processed directly.\n
Capability inventory: Execution of cluster-level CLI commands (flux, kubectl).\n
Sanitization: No validation or escaping of log content is performed before presentation or processing.

flux-troubleshooting