Scherlok

This server connects Claude to your data warehouse for automated data quality monitoring. It exposes tools for profiling tables, detecting anomalies like volume drops or schema drift, and checking freshness without writing rules or YAML configs. Works with Postgres, BigQuery, Snowflake, MySQL, and DuckDB. The MCP interface gives Claude read-only access to investigate, watch, list_tables, status, history, and check commands. Useful when you want an AI agent to spot data issues during development or review warehouse health on demand. Credentials stay server-side via environment variables, so the model never handles your database connection strings directly. Pairs well with dbt projects if you're already running transformations and want anomaly detection as a follow-up step.