How many times did you check out someone's branch from a pull request to then try things for yourself before performing the merge? Would it not be better if that test could be packaged and re-run for every future pull request automatically?

Or imagine you change your analysis code and everything is re-run from where it is needed and the paper draft is updated automatically?

I know that the tools exist (continuous integration, build checking, automated testing, bots that do your bidding) but I do not know the details.

If anyone know more about these, esp. using the CERN stack, it would be a great thing to learn about.