sota-kubernetes

Warn

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to access and search sensitive local configuration files, specifically within the ~/.kube/ directory, to identify static bearer tokens and ServiceAccount tokens. While this is a standard step in a Kubernetes security audit, accessing these paths involves handling highly sensitive authentication credentials.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including Kubernetes manifests, Helm charts, and cluster audit logs. This ingestion surface could potentially be exploited by malicious instructions embedded in these files. The skill includes mitigation strategies such as using helm template and kustomize build for pre-application review and implementing structured rollout plans for admission policies. The agent utilizes capabilities like kubectl, helm, and shell execution (grep, jq, talosctl) to perform these tasks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 6, 2026, 10:22 AM