The Agent Skills Directory

[SAFE]: No security issues were identified in this skill. The instructions and code samples follow standard practices for the PyHealth clinical deep learning library.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the pyhealth and torch libraries via the uv package manager from official registries. It also references synthetic clinical data hosted on Google Cloud Storage (googleapis.com). These resources are well-known services and appropriate for the skill's healthcare machine learning context.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external clinical datasets, which represents a theoretical ingestion surface for untrusted data. Ingestion points include classes like MIMIC3Dataset and eICUDataset. Boundary markers and sanitization are absent as the data processed consists of structured medical records and physiological signals intended for deep learning pipelines rather than natural language instructions aimed at the agent. Capabilities include dataset parsing, medical code mapping, and model training/inference.

pyhealth