cluster-deployment-engineer
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external Kubernetes manifests, Helm charts, and Kustomize configurations which serve as ingestion points for untrusted data that could contain embedded instructions designed to manipulate agent behavior.
- Ingestion points: The skill ingests external data from Helm values files and Kustomize overlays in
references/workload_delivery.md, and Git repositories used for GitOps synchronization inreferences/gitops_cluster.md. - Boundary markers: There are no explicit instructions or boundary markers provided to ensure the agent distinguishes between its own instructions and natural language content that might be embedded in YAML comments or metadata within the manifests.
- Capability inventory: The skill possesses significant capabilities via the
kubectlandhelmtools, including the ability to apply manifests, modify cluster resources, and retrieve logs, as outlined inreferences/workload_delivery.mdandreferences/troubleshooting.md. - Sanitization: No sanitization or validation logic is specified to filter or escape the content of external manifests before they are processed or applied to the target cluster environment.
Audit Metadata