About this project
Backer exists because I'd rather have backups automated than have to think about them. Every project on the server has a bit in its spec that says: when you set the project up, go to Backer and register anything that needs backing up. Usually that is a database. Sometimes it is other useful data that isn't part of a project. Backer then runs scheduled backups of whatever is on its list.
The offsite half is a separate job. Backer writes to a local directory on a schedule. Once a day an rclone job picks up everything in that directory and pushes it to the cloud. I kept those two concerns separate on purpose, so the backup tool only has to care about backups and cloud sync is someone else's problem.
The thing that works well about this setup is that it is properly fire and forget. I don't have to think about whether a new project is backed up, because the spec tells the Claude that created the project to register with Backer. By the time the project is running, its backups are already scheduled.
Screenshots
