The DuckLake Spec is so Simple, Even a Clanker Can Build One for Dataframes

The DuckLake team demonstrates the simplicity of the DuckLake v1.0 specification by using Claude Opus (AI) to autonomously build a Python dataframe reader/writer library called ducklake-dataframe. The library achieves read/write parity with DuckDB's DuckLake extension and supports Pandas, Polars, and PySpark with SQLite, Postgres, or DuckDB as catalog backends. The experiment highlights that DuckLake is a standalone spec — not tied to DuckDB at runtime — and is significantly simpler to implement than Apache Iceberg. Code examples show reading a DuckDB-written DuckLake from Pandas and writing back so DuckDB can read it. The library is published on PyPI but is considered a proof of concept, not production-ready.