Idea for discussion - make it work more like S6?

The uppermost similarity between your project and S6 is that acts as an init replacement.

Like S6-overlay, all of the configuration can be done on a volume that's not the root file system of the target system. You've chosen different naming conventions, but it might be possible to converge or adopt similar ones.

But S6 is not based entirely on shell scripts, so there's a difference that won't just go away. It's generally assumed that containers have networking configured and your system has to do that work itself. And S6 expects to persist for the lifetime of the container while yours quietly gets out of the way after boot. So they are not init replacements in quite the same sense.

But adopting the S6 configuration strategy might channel some adopters into following a known approach rather than rolling their own idiosyncratic naming system.