alterlab-pymc
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides legitimate Bayesian modeling workflows, templates, and documentation for the PyMC library. All operations, such as model fitting with MCMC or Variational Inference and diagnostic report generation, are consistent with the stated purpose of academic probabilistic programming.
- [SAFE]: File operations in the provided scripts (e.g., in
scripts/model_diagnostics.pyandassets/linear_regression_template.py) are limited to local persistence of analysis results using standard formats like NetCDF, CSV, and image files (PNG). No unauthorized data exfiltration or sensitive file access was detected. - [SAFE]: The documentation includes examples for model serialization using the
picklemodule inreferences/workflows.md. Whilepickle.loadcan be unsafe if used with untrusted data, it is presented here as a standard developer workflow for saving local progress. The skill correctly emphasizes and prioritizes the use of the safer NetCDF format via ArviZ for most persistence tasks.
Audit Metadata