jetson-video-recipe
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill invokes its core logic using
python3 -I, which enables isolated mode to ignore environment variables and user site-packages, effectively mitigating common path-based library injection attacks. - [SAFE]: The Python implementation in
scripts/recipes/recipe_model.pyuses a strict JSON loader that explicitly rejects duplicate keys and non-finite numbers (NaN/Infinity), which prevents potential logic bypasses via parser differentials. - [SAFE]: File creation in the recipe engine uses secure practices, including the
os.O_EXCLflag to prevent symlink attacks and setting file permissions to0o600(owner read/write only) to protect generated artifacts. - [SAFE]: The skill performs exhaustive validation of user-supplied intent against an internal catalog and strict schema constraints, ensuring that only authorized encoder parameters are processed.
Audit Metadata