This paper introduces Web2MexADL, a tool that can discover architectural documentation for web systems, by analyzing properties and relationships of their source code artifacts. This analysis is based on the use of probabilistic methods executed through machine learning tools. The resulting documentation includes an Architecture Description Language (ADL) document and a Scalable Vectors Graphic (SVG) file depicting an architecture view of the analyzed system. Finally, the resulting ADL document can be used to verify the maintainability of the system under analysis, by taking advantage of the MexADL approach.