tao-train-pointpillars
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected. The skill is consistent with its stated purpose of managing machine learning workflows.
- [EXTERNAL_DOWNLOADS]: The skill references official NVIDIA container images (e.g.,
nvcr.io/nvidia/tao/tao-toolkit) for its training and deployment actions, which is standard for this vendor's toolkit. - [INDIRECT_PROMPT_INJECTION]: The skill exposes a surface for indirect injection as it processes user-provided configuration parameters that are interpolated into shell commands.
- Ingestion points: YAML spec templates in the
references/directory and JSON schemas inschemas/. - Boundary markers: Absent in documentation, though the system uses structured configuration formats (YAML/JSON) to define job parameters.
- Capability inventory: The skill uses the
Bashtool to executepointpillarsCLI commands within a Docker container. - Sanitization: No explicit sanitization or validation of external content is documented within the skill itself.
- [NO_CODE]: This skill consists of documentation, configuration schemas, and spec templates; it does not include any standalone executable scripts or binaries.
Audit Metadata