pod-management
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous instructions for executing shell commands, including interactive shell access to containers (
kubectl exec -it), arbitrary command execution within pods, and port forwarding to the local network interface. - [PRIVILEGE_ESCALATION]: Includes commands for node-level debugging (
kubectl debug node/node-name), which can provide the agent with access to the underlying host node's filesystem and process namespace. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Untrusted data enters the agent context through cluster inspection commands such as
kubectl logs,kubectl describe, andkubectl get pod -o yamldocumented inSKILL.md. - Boundary markers: There are no explicit boundary markers or instructions for the agent to ignore potentially malicious content embedded in logs or pod specifications.
- Capability inventory: The skill provides powerful administrative capabilities including resource deletion (
kubectl delete), command execution (kubectl exec), and bidirectional file transfers (kubectl cp). - Sanitization: No sanitization, validation, or filtering of the data retrieved from the Kubernetes cluster is implemented or suggested.
Audit Metadata