lpbf-superres

Warn

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/sr.py employs torch.load(model, weights_only=False) within the predict and plate_compare functions. This facilitates arbitrary code execution via the unpickling process if a malicious model file is loaded.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted image and metadata inputs, creating a vulnerability surface. * Ingestion points: Image files via cv2.imread and metadata via np.load/json.load in scripts/sr.py. * Boundary markers: None. * Capability inventory: File writing (cv2.imwrite, np.save, torch.save) and complex tensor operations. * Sanitization: Inputs are treated as numerical data for processing without specific safety validation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 7, 2026, 06:02 AM