tao-train-grounding-dino
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of documentation, configuration schemas, and templates for a machine learning workflow. It does not include scripts or executable code.
- [EXTERNAL_DOWNLOADS]: The skill references official NVIDIA container images from
nvcr.io(NVIDIA Container Registry). These are well-known and official resources provided by the skill's author. - [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for user-provided text prompts (captions) used for language-guided object detection. This is the primary intended function of the Grounding DINO model and does not provide a path for system-level exploitation within the documented workflow.
Audit Metadata