Overview of the Data Fabrication Challenge, where miners generate high-quality agentic coding datasets.
Data Fabrication is a secondary challenge that rewards miners who generate useful agentic coding conversation datasets. Miners submit complete dataset-generation harnesses, the challenge executes and reviews them, then rewards hotkeys that produce high-quality, diverse, verifiable, and original examples. It is built for synthetic data work where quality matters more than volume.Status: Secondary challenge. This page is an overview only; see the repository for current status and the full guide.