tao-train-ocdnet
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation references official Docker images hosted on NVIDIA's container registry (nvcr.io) for model deployment.
- [COMMAND_EXECUTION]: Provides instructions for running TAO Toolkit CLI tools (ocdnet) and containerized environments for model training and deployment.
- [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection through image dataset processing. Ingestion points include dataset split folders (img/ and gt/) defined in SKILL.md. Boundary markers are established via the required directory structure. Capability inventory includes model training, evaluation, and quantization CLI tools. Sanitization is performed through configuration validation against the action-specific JSON schemas provided in the schemas/ directory.
Audit Metadata