troubleshooting
Pass
Audited by Gen Agent Trust Hub on Nov 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill employs the
kubectlCLI to perform administrative cluster operations. This includes viewing logs, describing resources, and modifying resource configurations (e.g.,kubectl set resources). It also useskubectl execandkubectl debugto execute shells and troubleshooting tools within the cluster environment. - [EXTERNAL_DOWNLOADS]: The skill references and uses several external container images for diagnostic purposes. These include the standard
busyboximage, the well-knownnicolaka/netshootnetworking toolkit, and official Kubernetes testing images fromgcr.io(a well-known registry). - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from potentially untrusted external sources within the Kubernetes cluster, creating a risk that malicious data could influence agent behavior.
- Ingestion points: Data is pulled from the cluster via
kubectl logs,kubectl get, andkubectl describecommands (SKILL.md). - Boundary markers: There are no explicit markers or instructions provided to the agent to distinguish between administrative output and potentially malicious content embedded in logs or resource metadata.
- Capability inventory: The skill has access to sensitive capabilities such as executing code in containers (
kubectl exec), debugging nodes (kubectl debug), and changing deployment specifications (kubectl set resources). - Sanitization: No sanitization or filtering logic is present to validate the content retrieved from the cluster before it is processed by the agent.
Audit Metadata