This is the MCP interface for GoldenFlow, a data normalization toolkit that fixes the usual mess: phone numbers in 15 formats, inconsistent dates, whitespace noise, and schema mismatches. Through Claude, you can invoke transforms on CSV, Excel, Parquet, S3 buckets, or databases. It exposes operations like auto-profiling columns, generating transform configs, mapping schemas between systems, and streaming large files in chunks. The zero-config mode auto-detects and applies safe fixes (E.164 phones, ISO dates, trimmed strings), while explicit configs let you define domain-specific rules. Available in Python and TypeScript with feature parity. Useful when you're cleaning data before deduplication or pipeline ingestion and want reproducible transforms instead of one-off scripts.
claude mcp add --transport stdio benzsevern-goldenflow uvx goldenflow