tao-train-single-step

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime intake (tao-launch-workflow) is driven by user-provided free-text parameters like train_dataset_uri, and the workflow does not specify any filtering/selection step that would prevent outsider-authored text from being ingested before model/action selection.