Currently, we have message passing rule derivations for (most?) of the nodes available internally in an Overleaf file. Some things can be improved, such as making it publicly available on a website like the book of statistical proofs, and restructuring it. The documentation for specific nodes can be marked features, which can be further divided into tasks (derivation, implementing in RxInfer and adding tests)