Coding the Backdoor Criterion to estimate the total effect of variables and selecting priors

So the model is structurally just some linear regressions. The closest example is for a moderation model Bayesian mediation analysis — PyMC3 3.11.5 documentation