gcp-gke-monitoring-observability
Warn
Audited by Gen Agent Trust Hub on Feb 24, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill references a sensitive file path used for service account authentication.\n
- Evidence: Reference to
/var/secrets/google/key.jsonin the Spring Cloud GCP configuration example withinreferences/advanced-configuration.md.\n- [COMMAND_EXECUTION]: The skill utilizes administrative commands to modify cluster state and interact with running workloads.\n - Evidence: Extensive use of
gcloud container clusters updateto modify logging and monitoring configurations inSKILL.mdandexamples/examples.md.\n - Evidence: Use of
kubectl execto perform health checks and metric queries within running pods inexamples/examples.md.\n- [PROMPT_INJECTION]: The skill processes untrusted external data from Kubernetes logs, creating a potential surface for indirect prompt injection.\n - Ingestion points: Data is ingested from container logs using
gcloud logging readandkubectl logsinSKILL.mdandexamples/examples.md.\n - Boundary markers: The skill does not implement delimiters or safety instructions when handling ingested log data.\n
- Capability inventory: The skill possesses the ability to update cluster settings, deploy resources, and execute commands within pods.\n
- Sanitization: Log data is read and processed without validation or sanitization of its content.
Audit Metadata