pyhealth
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches synthetic clinical datasets from a Google Cloud Storage bucket (storage.googleapis.com) and references the official PyTorch wheel registry (download.pytorch.org) for CUDA support. These are well-known and trusted infrastructure providers.
- [INDIRECT_PROMPT_INJECTION]: The skill processes clinical data files (CSV, ECG signals, imaging) from local or remote sources. While this data is technically untrusted, the skill follows standard medical machine learning practices for data ingestion and provides specific task definitions for processing these inputs.
- [COMMAND_EXECUTION]: Provides standard command-line instructions for managing Python environments using the uv tool, such as uv init, uv add, and uv run. These are utility commands for project setup and do not involve suspicious execution patterns.
Audit Metadata