The Agent Skills Directory

[SAFE] (SAFE): No malicious code, obfuscation, or suspicious patterns were found. The scripts use standard medical imaging and data processing libraries to perform their stated functions.
[Data Handling] (SAFE): The skill is designed for processing sensitive medical data (DICOM), and the included anonymization script specifically removes Protected Health Information (PHI) fields. No hardcoded credentials or external network requests were detected.
[Indirect Prompt Injection] (SAFE): While the skill processes untrusted external files (DICOM), it lacks exploitable capabilities like network access or shell execution that could be used for an attack.
Ingestion points: scripts/anonymize_dicom.py, scripts/dicom_to_image.py, and scripts/extract_metadata.py via pydicom.dcmread.
Boundary markers: Absent; the scripts process local files provided as command-line arguments.
Capability inventory: File system read and write operations only. No subprocess, eval, exec, or network libraries are utilized.
Sanitization: Relies on the pydicom library's standard parsing and attribute handling for DICOM datasets.

pydicom