I have some data (continuous) with heavy-tailed distribution. I found only Guassian Bayesian networks can be built in existing R and Python packages. I want to ask are there any tools to solve this problem. Thanks very much.
yup there are other ways. You can use Probabilistic Programming with PYMC3 e.g. (or Jags or Stan, but I have no experience with them), to build Bayesian networks with other non-normal probabilistic distributions.