k8s-kyverno-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides commands that modify the Kubernetes cluster state, specifically scaling deployments and deleting webhook configurations.
- Evidence:
kubectl scale deployment/kyverno-admission-controllerinSKILL.mdandkubectl delete validatingwebhookconfigurationinreferences/webhooks-and-controllers.md. - Mitigation: The skill includes strong instructional guards, labeling these as "Proposed fixes" that the agent should not run itself or must ask for explicit approval before executing.
- [INDIRECT_PROMPT_INJECTION]: The skill reads and processes potentially untrusted data from the Kubernetes cluster (logs, event messages, and resource manifests) which could contain malicious instructions.
- Ingestion points:
kubectl logs,kubectl events, andkubectl get -o yamlcommands used throughout all reference files. - Boundary markers: The skill instructs the agent to "Read first, ask before changing," but lacks technical delimiters to separate untrusted cluster data from instructions.
- Capability inventory: The agent has the ability to scale deployments (
kubectl scale) and delete configurations (kubectl delete) which could be abused if the agent follows instructions found in logs. - Sanitization: No specific sanitization or escaping of the cluster-returned data is performed before the agent processes it.
Audit Metadata