The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the download and use of security assessment tools from trusted, well-known organizations.
Fetches the kube-bench configuration and job manifest from Aqua Security's official GitHub repository.
Configures the Falco helm repository from the official Falcosecurity project for runtime threat detection.
[COMMAND_EXECUTION]: The workflow and audit script involve executing standard cloud and Kubernetes management tools.
Includes instructions for using eksctl, gcloud, az, kubectl, and helm to implement security controls like workload identity and network policies.
The scripts/agent.py script is designed to be executed manually to perform a read-only security audit of cluster resources via the official Kubernetes Python client.
[PROMPT_INJECTION]: An indirect prompt injection surface is present as the audit tool processes data from the Kubernetes cluster environment.
Ingestion points: The scripts/agent.py script reads metadata, labels, and specifications from Namespaces, Pods, and RBAC bindings via the Kubernetes API (file: scripts/agent.py).
Boundary markers: There are no explicit boundary markers used in the script to delineate between data and instructions within the ingested cluster metadata.
Capability inventory: The agent script has the capability to write its findings to a local JSON file (k8s_security_audit.json).
Sanitization: The script uses standard JSON serialization for the report, which provides inherent protection against data structure manipulation in the output.
[SAFE]: No malicious patterns or unauthorized behaviors were detected. The skill focuses on legitimate security hardening tasks and utilizes official libraries and trusted external resources.

securing-kubernetes-on-cloud