tao-train-nvdinov2
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: A comprehensive analysis of the skill's instructions, schemas, and reference documents found no evidence of malicious behavior, prompt injection, or unauthorized data access.
- [EXTERNAL_DOWNLOADS]: The skill pulls official TAO Toolkit container images from the trusted NVIDIA registry (nvcr.io). These are verified vendor resources.
- [COMMAND_EXECUTION]: The skill defines standard CLI commands for the nvdinov2 toolset, which are intended for use within containerized environments.
- [INDIRECT_PROMPT_INJECTION]: The skill involves processing external datasets and configuration parameters. 1. Ingestion points: Training and inference image directories specified in configuration files. 2. Boundary markers: None identified. 3. Capability inventory: Execution of nvdinov2 commands for model training, export, and inference within a container. 4. Sanitization: Requires dataset paths to be extracted directories rather than archives. This is standard functionality for machine learning training skills.
Audit Metadata