Build agent service

The application (service) I am talking about is an application that I am developping in cooperation with a psychologist, it's intended to time control the usage of computer time of kids that have game addictions. This part is mostly done and working. However. It was interesting to see how young kids seem to be able to bypass these things with ease. They kill or suspend processes, they experiment pulling out network cables etc.. any possible way to crash or stop the application. Initially I worked with a helper app that checks the process, however they manage to find it and kill it too. So I decided to work with a service that checks if the agent application is running under the profile of the user, if it's not, it has to be started again. This agent will log the user off when time is exceeded. Kind of the short version. Issues we had so far, user could not be identified, when logging on with RDP user, the user is recognised as system user. We do not want this running under system account. We have 2 possible ways to work. 1. The running service launches the application agent with the user credentials. 2. we let the service do the check and log off the specific user. If either of these 2 work I am happy. In short, I need an unkillable process. I know this sounds a lot like malware but it's not intended to be used for that reason, it would also have to be protected somehow, at some point, in order to avoid it being used for this reason. As I had no 100% solution to this issue so far I want to make sure that the dev that is creating this has knowledge of Windows OS and what issues they might encounter.