Very cool, it’s insane to me how frustrating the S3 UI is. My current annoyance is bulk downloading is hard, and seeing things like “how large is this bucket or specific key prefix in bytes”

Saw you using DuckDB at internal/handlers/api.go

Wondering if you’d considered letting queries operate at the bucket or key prefix level instead of requiring key to specific file? Or is that already supported?

I’ve liked being able to point DuckDB to a folder following Hive partition scheme, and then querying it just like Spark/Trino would over many files