managing-opa

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow uses kubectl get ... -o json and kubectl logs ... to ingest cluster-owned JSON fields and log lines (e.g., constraint templates, constraints, audit controller logs) into the agent’s LLM context; these are outsider-authored free text/data from other parties’ cluster workloads (not authored by the operating user).