tao-train-action-recognition
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were identified in the skill. The instructions and configurations align with standard practices for the NVIDIA TAO machine learning toolkit.
- [EXTERNAL_DOWNLOADS]: The skill references a Docker image
nvcr.io/nvidia/tao/tao-toolkit:7.3.0-pytfor executing training and inference tasks. The image is hosted on the official NVIDIA Container Registry, a well-known and trusted service provider. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates processing of video datasets for model training and evaluation (ingestion point: data volumes mounted to the container). While this involves ingesting external data, the capabilities are limited to tool-mediated machine learning operations within a containerized environment (capability inventory: Bash via Docker run). Standard boundary markers and sanitization for raw video data are not defined, as is typical for this class of machine learning skill.
Audit Metadata