GoldenCheck

This server brings zero-config data validation into Claude. Instead of writing rules upfront, it scans CSVs or tabular data to auto-discover validation issues like nulls, outliers, format problems, and drift from learned baselines. You can scan a dataset, get findings with severity and confidence scores, pin the ones you care about, and export them to YAML for CI validation. It exposes operations to profile data, create baselines, detect drift, and optionally enhance findings with LLM-generated context. Available as both stdio and HTTP transports. Note that the repo has moved to a monorepo at benzsevern/goldenmatch, so check there for active development. Useful when you need quick data quality checks without setting up heavy validation frameworks first.