I think it's an odd direction to go, focusing on a new "query language" and CLI tool rather than something that allows a git repo to be used as a storage backend in a normal RDBMS. Then you wouldn't have to implement the SQL grammar & logic, and you'd get a query optimizer for free. Using something like Postgres's foreign data wrappers maybe.

It doesn't look like this query language has much domain-specific syntax at all. It's just SQL with git as a storage backend.