pydicom
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill documents the use of the well-known pydicom Python library and its associated ecosystem (numpy, matplotlib, pillow) for medical image analysis. All recommended installations use standard package managers and point to official repositories.
- [INDIRECT_PROMPT_INJECTION]: The skill processes DICOM metadata which may originate from untrusted sources. This metadata (e.g., PatientName, StudyDescription) could potentially contain malicious instructions if processed directly by an LLM. 1. Ingestion points: DICOM file reading via pydicom.dcmread(). 2. Boundary markers: Absent. 3. Capability inventory: File system access (read/write) and image processing capabilities. 4. Sanitization: No specific sanitization or filtering of metadata for safety is described in the provided examples.
Audit Metadata