The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the histolab library via uv pip install and provides functionality to download sample pathology datasets (e.g., prostate, ovarian, breast tissue) from the TCGA (The Cancer Genome Atlas) project for demonstration purposes.
[DYNAMIC_EXECUTION]: The library's Lambda filter class is documented, which allows users to define and execute custom Python functions for image transformation and analysis (e.g., brightness adjustment, color channel extraction). This is a standard functional programming pattern within the histolab API.
[COMMAND_EXECUTION]: The skill demonstrates file system operations such as creating output directories for processed tiles and deleting low-quality tiles using Path.unlink() during quality control workflows.
[INDIRECT_PROMPT_INJECTION]: The skill processes Whole Slide Images (WSI) in formats like SVS, TIFF, and NDPI. As an ingestion point for external binary data, this presents a theoretical surface for indirect prompt injection if malicious data were embedded in slide metadata, though this is a standard risk for image processing applications.
Ingestion points: Slide object initialization with external WSI file paths in SKILL.md and references/slide_management.md.
Boundary markers: None present for binary image data.
Capability inventory: The skill performs extensive file system writes, including saving thumbnails, extracting thousands of image tiles, and generating CSV/PDF reports.
Sanitization: Relies on standard libraries like OpenSlide, Pillow, and OpenCV for image handling and processing.

histolab