Docs

Introduction

Data Shepherd is a data quality monitoring and alerting app for Snowflake. You point it at tables you already have, set rules against them, and it records what it finds and tells you when something is wrong.

It is a Snowflake Native App, so it installs into your own Snowflake account and runs there. Your data is read in place and never leaves your account.

What it checks

Twenty-five built-in rules ship in five families:

  • Core checks - completeness, uniqueness, value range, data freshness and referential integrity, plus profiling and an overall quality score.
  • Validation checks - allowed-value domains, conditional completeness, numeric-coercion detection and fuzzy duplicate detection.
  • Historical checks - distribution drift, period comparison, trend analysis, seasonal validation and late-data detection, each compared against your own history.
  • Pattern checks - regular-expression patterns, pattern comparison across segments, time-series pattern detection and density clustering.
  • Schema-drift check - alerts when a column is added, dropped or retyped on a table you have connected.

Every rule is read-only. None of them writes to, updates or deletes a row in the data you monitor.

Why in-account matters

  • Zero data egress. Rules run inside your account and write their results to the app’s own tables there. The only things that can leave are the email and webhook notifications you choose to set up.
  • Access you can see. The app’s only access to your data is SELECT on the tables you connect, and you can disconnect any of them at any time.
  • Nothing new to host. The app opens in Snowsight, next to the rest of your Snowflake work. There is no separate service to run or sign in to.

Where to go next

← Back to docs