SchemaSleuth

SchemaSleuth is a web app (traditional + AI) that monitors analytics and data warehouse tables for silent schema drift and contract violations. It connects to Snowflake/BigQuery/Postgres, learns expected column types, null rates, cardinality ranges, and key constraints, then alerts when upstream changes would break dbt models, BI dashboards, or ML feature pipelines. The AI layer summarizes what changed, likely root causes (e.g., new enum value, widened type, renamed field), and suggests safe fixes (dbt model patch, view shim, or contract update). It also generates lightweight “data contracts” from observed behavior and lets teams approve changes via pull-request style workflows. This is not a full observability suite; it’s a focused guardrail for teams that keep getting surprised by breaking changes. Expect some false positives early; tuning and ownership workflows are part of the product.