tao-train-foundation-stereo
Pass
Audited by Gen Agent Trust Hub on Jun 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references pretrained model weights hosted on Hugging Face (
huggingface.co). This is a standard practice for initializing deep learning models and utilizes a well-known service in the AI community. - [COMMAND_EXECUTION]: The skill provides instructions for running the TAO Toolkit
depth_netCLI within Docker containers. These instructions follow security best practices, such as running containers with the host user's UID/GID ($(id -u):$(id -g)) to avoid root privileges inside the container and mounting data volumes as read-only where appropriate. - [SAFE]: No instances of prompt injection, data exfiltration, or obfuscation were detected. The skill's functionality is strictly aligned with its stated purpose of providing a model training and deployment workflow.
Audit Metadata