The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The skill instructions include high-privilege operations such as kubectl exec, kubectl debug, and cluster resource modification (kubectl set resources). While necessary for troubleshooting, these commands grant the agent broad access to the environment.
[EXTERNAL_DOWNLOADS] (LOW): The skill references external container images for debugging purposes, including nicolaka/netshoot and busybox. While gcr.io is a trusted source, the use of community-maintained images like netshoot introduces a dependency on external, third-party content.
[PROMPT_INJECTION] (LOW): The skill is vulnerable to indirect prompt injection (Category 8).
Ingestion points: The agent is instructed to read untrusted data via kubectl logs, kubectl get events, and kubectl describe pod.
Boundary markers: None present in the instructions to help the agent distinguish between log content and system instructions.
Capability inventory: The agent has access to powerful tools like kubectl exec, kubectl debug, and kubectl run.
Sanitization: No sanitization or validation of the retrieved log/event data is suggested before the agent processes it.

troubleshooting