benseverndev-oss/goldenmatch — reverse-engineered prompt

Build me a polished data quality and record matching toolkit that feels zero config for normal use. I want to be able to point it at a CSV or database table and have it profile the data, flag obvious quality issues, standardize things like names, phones, dates, and addresses, then dedupe and link records into clusters and produce golden records. It should support exact matches, fuzzy matches, probabilistic matching, optional privacy preserving matching, and an identity graph so matched entities can stay stable across runs.

Please also include a simple web workbench where I can inspect matched pairs and clusters, see field by field differences, read a short plain English explanation for why two records matched, and review or correct decisions. I want reports I can export, a pipeline mode that chains the steps together, and defaults that self check instead of silently doing the wrong thing. It should work well on a laptop for small files but be able to scale to very large datasets too. If anything is unclear, look up the current docs online and make sensible choices.