The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/preflight_series.py) to perform DICOM header analysis. The execution is performed according to standard patterns defined in the manifest.
[COMMAND_EXECUTION]: The test suite in tests/test_preflight_series.py includes a subprocess.run call to execute a local fixture generation script. This is a standard development practice and is implemented safely using argument lists and the current system executable.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8) as it reads metadata from external DICOM files. This is documented and inherent to the utility's purpose. Evidence chain:
Ingestion point: scripts/preflight_series.py reads tags like SeriesDescription from files in a user-provided directory.
Boundary markers: The results are returned in a structured JSON format.
Capability inventory: The skill is permitted to use the Bash tool.
Sanitization: Extracted strings are not sanitized for potential instruction patterns, which is a standard limitation for this use case and does not represent a malicious finding.

dicom-series-preflight