KIMMDY is a software extending the possibilities of established Molecular Dynamic techniques to chemical reactions. Traditional Molecular Dynamics (MD) simulations can not model chemical reactions, although they would be crucial to understand many processes like aging of skin and tendons. KIMMDY enables studying such systems computationally by combining classical MD simulations with kinetic Monte Carlo (kMC) steps to model chemical reactions. The kMC steps depend on reaction rates, which are obtained using machine learning-, or physical models.This talk highlights the creation process and architecture of KIMMDY. At its core, it is a workflow manager written in python, handling and moving data between machine learning models, physical models, and existing MD software. KIMMDY is build in a modular way, with many parts implemented in separate plugins. To orchestrate dependencies between them, and ensure compatibility, we use uv workspaces. Since we welcome future contributions from outside parties, we aim to make the barrier of entry as low as possible. To that end, we have ensured that modern editors can recognize our input file format and provide help text, and perform type checking through the use of json schemas. Furthermore, we provide documentation and tutorials rendered by Quarto to include executable code in the webpage creation process, generating visually pleasing, up-to-date documentation.