k8s-rbac-troubleshooting

Pass

Audited by Gen Agent Trust Hub on Nov 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: No malicious patterns detected. The skill provides standard administrative procedures for Kubernetes RBAC debugging. It includes robust safety markers instructing the agent to never read secrets or tokens and to always seek user approval before executing any state-changing commands.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes external, potentially untrusted data from cluster logs and events.\n
  • Ingestion points: kubectl logs and kubectl get events as referenced in SKILL.md.\n
  • Boundary markers: Absent; the skill relies on standard output parsing within the agent's context.\n
  • Capability inventory: Limited to kubectl investigation tools (get, describe, logs); state-changing actions like kubectl apply are delegated to the user or require explicit approval.\n
  • Sanitization: Absent.\n Despite this surface, the risk is mitigated by the 'human-in-the-loop' requirement for all proposed fixes and the narrow scope of investigation.\n- [COMMAND_EXECUTION]: The skill uses kubectl to perform cluster diagnostics. This is consistent with its stated purpose and does not involve unauthorized privilege escalation, hidden network operations, or hardcoded credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 8, 2026, 10:58 PM