I've been thinking how we could solve the following problems that we sometimes encounter: - failing database migrations - breaking changes in Rust-DLC I realised that it might be a quite simple solution for this, by storing the databases from e2e test runs. This way we would know whether we're about to introduce a regression in *maker*, *app*, or *coordinator*. unsolved problem is how to deal with the fact when we decide that it's OK to break things - perhaps manually clearing the action cache via the CLI or web interface would be enough in such situation.