surf-kubectl-troubleshoot

Warn

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill accesses the sensitive file path ~/.config/sops/age/keys.txt to diagnose SOPS decryption failures. Accessing private cryptographic keys, even for troubleshooting, increases the risk of credential exposure.
  • [DATA_EXFILTRATION]: The execution of kubectl get secrets, kubectl describe pod, and kubectl logs can expose sensitive information in the agent's context, including platform-provisioned database credentials, application secrets, and environment variables.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted output from cluster resources and external tools which could be manipulated by an attacker to influence the agent's logic.
  • Ingestion points: Output from kubectl logs, kubectl describe, kubectl get events, and git status (as specified in SKILL.md).
  • Boundary markers: None identified; the instructions do not implement markers or warnings to disregard instructions embedded in the analyzed logs or resource metadata.
  • Capability inventory: The skill has access to shell execution, resource annotation via kubectl, file system permission changes (chmod), and local script execution.
  • Sanitization: There is no evidence of sanitization or validation of the data retrieved from the Kubernetes API or logs before it is interpreted by the agent.
  • [COMMAND_EXECUTION]: The skill instructs the agent to run various shell commands, including kubectl, git, sops, and bash. It specifically references local scripts such as scripts/sops-doctor.sh and uses kubectl annotate to perform reconciliation tasks.
  • [DYNAMIC_EXECUTION]: The skill generates commands by interpolating user-provided inputs like <name>, <ns>, and <app> into shell templates. This pattern presents a potential for command injection if the inputs are derived from untrusted sources and not properly handled by the platform.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 31, 2026, 11:08 AM