Once the mbx script produces results, it is your responsibility to keep track of them. Early in a project, you may wish to regenerate images regularly, and not save the results permanently (for instance, you may not want to track them under version control). In this case, a build script is very useful. For a mature project, you may only regenerate images you know have changed, or when you create a new edition. And you may wish to include them with your source files, under revision control, for your readers to use.
More critically, you need to place items generated by the mbx script where the rest of your output can find them. For example, by default HTML output expects images to be in a subdirectory named images. (This default can be changed, though it seems almost nobody does.)