We want an encrypted backup app fully supporting BIP39/SLIP39 (beyond what SeedVault supports) which backs up app/OS data to your home directory, a storage drive or via a device-to-device transfer. It doesn't need to initially support backing up home directory files since they can already be backed up without it. t will always use device-to-device mode meaning that allowBackup="false"
and the legacy backup file exclusion system will be completely ignored since they only apply to cloud backups. One of the main issues with SeedVault is that it's considered to be a cloud backup. It should also support choosing which apps get backed up, and if and when backing up the home directory is supported which parts of that are included.
It doesn't need to support SAF or cloud storage since people can simply grant use a cloud storage app and tell it to sync the backup directory. GrapheneOS supports Storage Scopes now which means users never need to give access to files in their home directory to use apps and therefore having things in your home directory is a much better approach than it ever was in the past since there's no need to give apps access to everything. We should still get more users doing this as intended, but we don't need to focus on working around the storage model anymore.
We would focus on making a robust, highly usable app providing the bare minimum and then we can improve it from there. Look at our app repository client for the most recent example of an app we built ourselves without a legacy codebase we need to improve upon. It's a very good app and supports a lot of modern features not included in other app repository clients including split APK support, selection of best match release based on min API level / architecture, fs-verity, dependency management with circular dependency support, atomic multi-package installation for groups of packages with interdependencies, release channels, parallel downloads/installs, automatic update checks, idle-only fully automatic updates to avoid respawning apps that are being actively used including when it isn't a privileged app, compressed app downloads, resuming interrupted downloads, proper metadata signing with key rotation, TLS key pinning, etc. It also has a pretty good user interface. It's actually a very good app, which we built from scratch and expanded based on our needs. We can do the same thing with a backup app but we need more developers. The app repository client is a shining example of what we're capable of doing now when we apply significant resources to a task like this for enough time. Our other apps haven't gotten the same level of attention yet, but they could. Camera app is in a fairly similar state as the app repository client before the recent rewrite to greatly improve it. Our other apps are much older and aren't as good (note: not including the badly aged AOSP apps as ours).
We simply need to do this ourselves. No one else is going to build what we need for us. They do not share our approach or goals, they don't know how to make great software and they're incredibly hostile towards not only us but also our users. They lie to our users when they report issues to blame them on us. There's no path forward beyond replacing the app. We can continue updating it but it's not getting less awful.