The Agent Skills Directory

[SAFE]: The skill serves as a guide for Kubernetes security auditing. It describes methodologies for identifying exposed components such as the API server, kubelet, and etcd, and explains how to test for RBAC misconfigurations and container breakout scenarios.
[REMOTE_CODE_EXECUTION]: References standard security tools such as kubectl, kube-hunter, and peirates. These are well-known utilities in the security community used for cluster assessment.
[DATA_EXPOSURE]: Lists sensitive file paths (e.g., /var/run/secrets/kubernetes.io/serviceaccount/) and commands for accessing cluster information (e.g., kubectl get secrets -A). In the context of the 'pentest' and 'bugbounty' modes specified in the frontmatter, these are identified as targets for authorized testing rather than evidence of malicious behavior within the skill itself.
[INDIRECT_PROMPT_INJECTION]: While the skill involves reading data from external Kubernetes environments, it does not demonstrate unsafe interpolation of that data or attempts to influence the agent's behavior through the processed content.

cloud-kubernetes