azure-ai-anomalydetector-java

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

• Third-party content exposure detected (high risk: 1.00). The skill's Multivariate Model Training and Multivariate Batch Inference examples explicitly set a dataSource to arbitrary blob URLs (e.g., "https://storage.blob.core.windows.net/container/data.zip?sasToken") which the agent would fetch and ingest for training/inference, allowing untrusted third-party content to influence model behavior.