Overview

droughty — adjective, drought·i·er, drought·i·est. dry.

Droughty helps keep your analytics workflow DRY — Don’t Repeat Yourself.

The problem

Analytics engineering is full of mechanical, repetitive work that nobody loves doing but everyone has to do:

  • Your Looker views fall behind your warehouse because manually updating 40 views when a column changes isn’t anyone’s idea of a good time.

  • Your dbt schema.yml tests are either missing or out of date because writing them from scratch is tedious.

  • Your ERD diagram on Confluence hasn’t been updated since the project kicked off.

  • Your Cube schema drifted from the warehouse two sprints ago and nobody noticed.

  • Half your warehouse fields have no description, so new analysts waste hours asking “what does this column mean?”

The real cost isn’t just time — it’s the bugs and confusion that creep in when your semantic layer doesn’t match your warehouse. Dimensions in Looker that reference columns that no longer exist. dbt tests that were never written, so a broken primary key makes it to production. ERDs that mislead rather than inform.

The solution

Droughty reads your data warehouse’s INFORMATION_SCHEMA and generates these files automatically. Your warehouse schema becomes the single source of truth, and every tool downstream stays in sync with it.

droughty lookml    # → LookML views, explores, measures
droughty dbt       # → schema.yml with tests for every table
droughty dbml      # → ERD with all PK/FK relationships
droughty cube      # → Cube schema with dimensions and measures
droughty docs      # → AI-generated field descriptions
droughty qa        # → LLM-powered data quality validation
droughty stage     # → dbt staging models from source tables

Run these commands as part of your development workflow and the mechanical stuff takes care of itself — so you can focus on the work that actually requires thinking.

What droughty generates

Command

Why it exists

What it generates

Target tool

lookml

Looker views drift from the warehouse every time a schema changes. Droughty keeps them in sync automatically.

LookML views, explores & measures

Looker

dbt

Writing schema.yml tests is the task that always gets skipped. Droughty generates baseline coverage for every table instantly.

schema.yml with uniqueness, not-null & validity tests

dbt

dbml

No one updates the ERD. Droughty generates it from the live warehouse, so it’s always accurate.

DBML ERD with PK/FK relationships

dbdocs / dbdiagram.io

cube

Same as LookML — schema drift is silent and expensive. Droughty keeps Cube in sync.

Cube dimensions, measures & joins

Cube.js

docs

Undocumented fields kill self-service analytics. AI gives you a first-pass description for every field in seconds.

Markdown field descriptions via OpenAI

Any

qa

Data issues are usually found by stakeholders, not the team that built the pipeline. Droughty validates your data proactively.

LangGraph QA agent + Mermaid DAG

Any

stage

Writing boilerplate staging SQL for new sources is pure mechanical work. Droughty does it from the source schema.

dbt staging models and sources.yml

dbt + BigQuery

Supported warehouses

  • Google BigQuery

  • Snowflake