The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill provides legitimate guidance for researchers.
[INDIRECT_PROMPT_INJECTION]: The skill provides functions to process image metadata and text strings (titles/legends) for validation against journal standards. \n
Ingestion points: image_path in check_nejm_resolution and validate_nejm_figure; title and legend in check_clinical_image_text. \n
Capability inventory: Metadata reading via the Pillow library. No network, file-write, or shell execution capabilities are present across any scripts. \n
Boundary markers: Not present, as the code is designed for local validation and formatting. \n
Sanitization: The skill performs standard metadata extraction and string length checks.

nejm-figure-guide