the idea is to make deployments repeatable, so while we may allow sshing in the future, it will be for diagnostic/experimentation reasons, not for altering the device state, so I think you understand the problem we're solving pretty well.

We keep adding devices and will soon release a guide on how users can add their own devices to the mix. That said, the primary determinant on whether we can support a device is whether a yocto/openembedded BSP exists and is relatively modern (uses a kernel above 3.8). If that exists, it's almost certain that resin support will be relatively easy. Happy to chat more, email in profile.