If anyone interested, here is a superclass of Stream that relies on pydantic for schema discovery and an extension of this to support multithread ingestion: • https://github.com/prevostc/beefy-databarn/blob/main/extract/tap-beefy-databarn/tap_beefy_databarn/singer/pydantic_dataclass_stream.py • https://github.com/prevostc/beefy-databarn/blob/main/extract/tap-beefy-databarn/tap_beefy_databarn/singer/pydantic_multithread_stream.py Feedback welcome, I'm not primarily a python dev