Skip to content

dbdocs

dbdocs logo

An alternative dbt docs site — catalog + ERD + column-level lineage.

Turn your dbt artifacts into a single self-contained index.html: a browsable catalog, an interactive lineage DAG and ERD, and column-level lineage derived straight from your compiled SQL. No server, no database, no build step — just a file you can open or host anywhere.

🚀 Try the live demo Quickstart

The demo is a real dbdocs site built from the jaffle_shop dbt project — poke around the catalog, the lineage DAG, the ERD, and column-level lineage.

What you get

dbt's own docs are great until you want lineage at the column level — that's the gap dbdocs fills.

  • Catalog navigation grouped by database and schema, with client-side search (no backend).
  • Per-model detail — columns (type / tags / description), compiled and raw SQL, and the macros each model resolves.
  • Interactive graphs — a node-level lineage DAG and an ERD, both built on React Flow: pan / zoom / minimap, automatic dagre layout, filter-and-focus, and deep-links to any node.
  • Column-level lineage traced from each model's compiled SQL via sqlglot.
  • Dark / light theme.
  • Versioned deploy with a built-in version dropdown — no mike, no plugins.

Installation

Requires Python 3.10+

dbdocs leans on the dbt artifact parser and sqlglot, both of which long ago moved past Python 3.9 — so we did too. Upgrading your interpreter is the way forward (it's worth it).

pip install dbdocs --upgrade Successfully installed dbdocs restart ↻

Verify installation:

dbdocs --version

Quickstart

First, produce dbt artifacts in your dbt project (the bit dbdocs reads):

dbt docs generate           # writes target/manifest.json + target/catalog.json

Then generate, serve, and open the site:

dbdocs generate             # builds ./site/index.html with all data baked in
dbdocs serve                # static http server on http://127.0.0.1:8000

Head to the Quickstart guide for the full walkthrough.

Contributing

Contributions are welcome — bugs, features, docs, typos. See the Contributing Guide.

If dbdocs saves you some clicks, consider buying me a coffee.

buy me a coffee

**Made with ❤️ by Dat Nguyen**