This is not a fully automatic model builder. For standard models, use the brms Model Builder β there the generated code runs directly.
This tool is for rare, advanced building blocks that don't fit into the standard model builder β multivariate models, random-effects correlation structures, splines/Gaussian process, nonlinear formulas, autocorrelation, multi-membership, censoring/truncation, hurdle and ordinal families, measurement error and missing data, among others. It assumes experience with brms.
The sidebar on the left builds the code at the top right live. Every documentation card further down has a "β Apply directly" button that applies that setting straight into the model. Each card also has a collapsible "π§ͺ Data simulation" field with R code that generates a small test dataset for exactly that building block. The code in that field can be freely edited; once it is, changes in the sidebar no longer overwrite it automatically β "π Regenerate" deliberately rebuilds it from the current sidebar state, "βΊ Reset" also discards the sidebar state and starts over with an empty model.
Priors are deliberately not guessed β the generated code therefore contains no automatically generated priors <- c(...) block, but instead calls get_prior(), which shows which classes/coefficients brms expects for that particular model.
The "β Formal check" only checks for structural oddities (parentheses, missing arguments, tag consistency) β it is not a substitute for an actual test run in R (e.g. make_stancode() or a few iterations with sample_prior="only").