tao-train-mask-grounding-dino

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

• Third-party content exposure detected (medium risk: 0.30). Required runtime reads user-supplied inference prompt captions (free text) via dataset.infer_data_sources.captions into the model’s inference job.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's TAO Deploy examples run docker with the remote container image nvcr.io/nvidia/tao/tao-toolkit:6.26.3-deploy which would be pulled and executed at runtime (remote code execution dependency for deploy actions).