The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external DICOM files which could contain malicious instructions embedded in metadata fields (e.g., PatientName, StudyDescription).
Ingestion points: Uses pydicom.dcmread() to load file contents in SKILL.md, scripts/anonymize_dicom.py, scripts/dicom_to_image.py, and scripts/extract_metadata.py.
Boundary markers: None; the instructions do not include delimiters or warnings to ignore instructions found within the data.
Capability inventory: The skill can read and write to the filesystem and process pixel data. No network capabilities or dynamic code execution surfaces were found.
Sanitization: Medical metadata is extracted and displayed or saved to new files without validation or escaping of the string content.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install several third-party libraries via uv pip.
Dependencies: Recommends pydicom, pillow, numpy, matplotlib, pylibjpeg, pylibjpeg-libjpeg, pylibjpeg-openjpeg, and python-gdcm from standard package registries.
[COMMAND_EXECUTION]: The skill provides and demonstrates the use of local Python scripts to perform tasks.
Evidence: Instructions include execution of scripts/anonymize_dicom.py, scripts/dicom_to_image.py, and scripts/extract_metadata.py via the command line.

pydicom