dicom-series-to-volume

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). At runtime, the skill reads DICOM files from the caller-provided dicom_dir via scripts/series_to_volume.py::_read_series() using pydicom.dcmread(), then ingests DICOM header fields (e.g., ImageOrientationPatient, ImagePositionPatient, PixelSpacing, RescaleSlope/Intercept, and Study/Series* strings) into the JSON it prints—so any outsider-authored DICOM directory contents become LLM-readable free text in result_json (e.g., dicom_metadata, input_dir).