getitune-optimizing-a-model
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external datasets for model calibration during the optimization workflow, which represents a potential attack surface if the dataset files contain malicious content.
- Ingestion points: The
dataparameter in thecreate_enginefunction (e.g.,ov_engine = create_engine(data="/path/to/dataset")). - Boundary markers: Not specified.
- Capability inventory: Local file system access for reading datasets and writing optimized model files.
- Sanitization: Not mentioned in the skill instructions.
- [EXTERNAL_DOWNLOADS]: The skill references the official OpenVINO NNCF repository on GitHub for documentation purposes. This is an informational reference to a well-known toolkit.
Audit Metadata