Commit messages are important parts of the software development lifecycle, allowing developers to quickly understand changes made to a codebase. However, many developers do not write good quality commit messages. Therefore, a current area of research is the ability to automatically generate commit messages based on a set of changes. This project gathers existing GitHub commit messages and filters them based on a set of quality criteria, producing higher quality datasets for researchers to train machine learning models for automatically generating commit messages. Furthermore, this project accomplishes this in a scalable and customizable way, allowing for various projects to make use of the project as required.