securing-kubernetes-on-cloud
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references and downloads security configuration and tooling from established, trusted industry sources. These include fetching a CIS benchmark job from Aqua Security's official GitHub repository and runtime security charts from the Falco Security project's Helm registry. These operations are performed as part of the intended security auditing and monitoring functionality.\n- [COMMAND_EXECUTION]: Instructions are provided to use standard cloud CLI tools (eksctl, gcloud, az) and Kubernetes utilities (kubectl, helm) to configure the environment and deploy security resources. The included Python script performs local file operations to save audit reports but does not execute arbitrary shell commands or external scripts.\n- [INDIRECT_PROMPT_INJECTION]: The auditing script (scripts/agent.py) reads configuration data (labels, pod specs, RBAC bindings) directly from the Kubernetes API. \n
- Ingestion points: Data is ingested through API calls such as list_namespace and list_pod_for_all_namespaces in scripts/agent.py.\n
- Boundary markers: None; the script processes cluster metadata directly.\n
- Capability inventory: The script's capabilities are limited to reading cluster metadata, printing to stdout, and writing to a local JSON file.\n
- Sanitization: No specific sanitization is performed on the ingested metadata before it is written to the report.\n
- Assessment: While this represents an ingestion surface for untrusted cluster data, it is a fundamental requirement for the skill's purpose. The behavior is safe as the data is not used to trigger further autonomous agent actions or command execution.
Audit Metadata