Docglow

Docglow exists because thousands of teams run dbt Core without dbt Cloud, and the built-in dbt docs generate && dbt docs serve produces a documentation site that’s functional but limited — no full-text search worth relying on, no health scoring, and a lineage view that doesn’t scale well to large projects. Docglow generates a richer static documentation site from the same dbt project metadata.

It adds a lineage explorer for tracing model dependencies visually, health scoring to surface data quality or documentation gaps at a glance, and full-text search across models, columns, and descriptions — the kind of features teams would otherwise only get by paying for dbt Cloud.

MIT licensed and distributed as a PyPI package, Docglow generates a static site (a live demo is hosted separately), meaning the output can be deployed anywhere static files can be served, without depending on a running service.

What You Get

• A richer static documentation site generated from dbt Core project metadata
• A lineage explorer for visually tracing model dependencies at scale
• Health scoring to surface documentation and data-quality gaps across models
• Full-text search across models, columns, and descriptions

Common Use Cases

• Generating a usable, searchable documentation site for a dbt Core project without paying for dbt Cloud
• Visualizing model lineage in a large dbt project where the built-in dbt docs lineage graph becomes unwieldy
• Auditing documentation coverage and data quality gaps across dbt models via health scores
• Publishing dbt project documentation as a static site deployable to any static hosting provider

Under The Hood

Architecture Docglow reads dbt Core’s project metadata (manifest and catalog artifacts) the same way dbt docs generate does, but renders it into a richer static site with added features — lineage exploration, health scoring, and search — layered on top rather than replacing dbt’s own metadata generation. Because the output is a static site, Docglow’s own runtime concerns end at generation time; the resulting site has no server-side dependency.

Tech Stack Python, distributed via PyPI (pip install docglow), consuming dbt Core’s standard project artifacts as input and producing a static HTML/JS documentation site as output. CI is configured via GitHub Actions.

Code Quality The project publishes versioned releases to PyPI with download tracking, runs CI on every change, and maintains a changelog — signals of a project built for external, ongoing adoption rather than an internal one-off tool, backed by very active recent development.

What Makes It Unique Rather than building a new dbt-adjacent tool from scratch, Docglow specifically targets the exact gap between dbt Core’s built-in docs and dbt Cloud’s paid documentation features — lineage, health scoring, and search — giving dbt Core users functionality otherwise reserved for a paid tier, as a static site generator rather than a hosted service.

Self-Hosting

Licensing Model MIT licensed — fully open source with no license key.

Self-Hosting Restrictions Not applicable; Docglow generates a static site you host anywhere, with no separate server component to self-host.

License Key Required No.