First off, this is very cool! So don't let my comments dissuade you...

But IME serialization tends to fall into two boats:

1. I want something akin to a config that's human readable but very simple and I don't want to think about strong typing 2. I'm serializing data and I want schemas, efficiency, etc.

I feel like (1) is pretty well handled by TOML and JSON and (2) is pretty well handled by flat/proto buffers, thrift/avro, capn' proto, etc.

I guess I'm wondering where you see this being used.