Several MCMC routines are included with a common interface: Metropolis-Hastings, slice sampling, and Hamiltonian Monte Carlo. All you need to do is specify a function that computes an unnormalized density (e.g. a posterior distribution). For HMC, you’ll need to set the attribute of this value to be the gradient.
This is a similar effort to Radford Neal’s GRIMS scripts. This package is less general and flexible, but so far I have found it easier to debug and extend.