Both Metabase and Elasticsearch are secondary indexes - you’d typically configure them so that changes in the “main” database are pushed to them.
Be very wary of this practice - it’s essentially treating the entire hundreds-of-tables game database as a public API. If the game doesn’t provide a real API, I recommend you write a narrowly-focused one that writes to the things you need and use that instead of sprinkling random INSERTs throughout your code.
IMO this friction is a sign that you’re working against the grain of the tool; Ecto is primarily designed to handle building SQL queries that are known at compile time, and that’s the opposite of what you’re building.
I’m curious about the exact sort of users you’re envisioning for this; there’s a very narrow window of skill between “can understand a schema with hundreds of tables well enough to query it” and “can write SQL”, and the abstraction you’ve described doesn’t remove very much complexity.
