pymc

pymc is an agent skill for building and checking Bayesian models with PyMC, including hierarchical models, NUTS sampling with MCMC, variational inference, predictions with mutable data, and posterior predictive checks. It targets PyMC 6.3.2, PyTensor 3.3.2 and ArviZ 1.3.0. These are local Python interfaces. No service endpoint, credentials or remote inference is needed. The native tests use the PyMC NUTS sampler on small synthetic models. Optional nutpie and JAX samplers, sequential Monte Carlo, and convergence at production scale were not validated by those tests. For a scientific run the skill asks you to record the complete resolved environment, the backend, the random seeds, the preprocessing and the data provenance. The review notes an environment detail. On macOS, a C linker error could be avoided by disabling PyTensor C compilation for the process, and the Python library workflows passed under that setting. That does not show that the C backend works on the same host. Use it when you are fitting a Bayesian model and want the sampling diagnostics and the posterior checks handled properly.…