Nowadays, cloud services rely extensively on the use of virtual machines to enforce security by isolation. However, hardware trojan attacks break this assumption. Within these attacks, cache side-channel attacks such as Spectre and Meltdown are the focus of this work. In this project, we develop a set of tools to generate a dataset; and a dataset that will allow the use of Machine Learning techniques to detect Spectre and Meltdown attacks (i.e. using a cache side-channel). When released, this dataset will enable researchers to compare their ML-based detection proposals based on the same dataset (which is not currently the case). Also, it eliminates the need of an infected computer to generate the attacks and the corresponding dataset for subsequent research studies.