The Agent Skills Directory

[SAFE]: The skill documents the use of scvi-tools, a widely recognized open-source library for single-cell genomics. All code examples follow standard API patterns for the library.
[EXTERNAL_DOWNLOADS]: The skill suggests installing 'scvi-tools' via 'uv pip install'. This is a standard and expected procedure for using the documented library.
[DATA_EXPOSURE]: The skill demonstrates standard data handling practices, such as loading datasets from local files (e.g., .h5ad) and saving results locally. No patterns of sensitive data exfiltration or hardcoded credentials were found.
[PROMPT_INJECTION]: Instructions use natural educational language to guide the user through analysis workflows. No attempts to override safety guidelines or bypass agent constraints were identified.
[DYNAMIC_EXECUTION]: While the skill involves training deep learning models, it uses standard library calls for probabilistic modeling and does not employ unsafe dynamic execution methods like eval or exec on untrusted inputs.

scvi-tools