Генерация синтетических датасетов с Data Designer

★ 7.6 · content

data-designer is a Claude Code skill that builds synthetic datasets using the Data Designer library. It supports two modes: Interactive (default), where it asks clarifying questions, and Autopilot, triggered when the user says things like "just build it" or "make reasonable assumptions." The output is a Python file containing a `load_config_builder()` function that returns a `DataDesignerConfigBuilder`, optionally including sampler columns with typed params, LLM judge columns with Pydantic schemas, custom column generators, and PEP 723 inline dependency metadata. It's aimed at developers and ML practitioners who need to quickly produce training data, test datasets, or a prototype data-generation pipeline.