Tracing dependencies across 50+ DWH objects manually takes time, and documentation is outdated the moment you finish. I stopped after 20.
Existing lineage tools struggle with complex stored procedures, are expensive, or cloud-dependent. So I built my own.
The Build
Used Claude Code. More time-consuming than expected: planning, building, changing, and lots of testing — especially keeping parsing rules stable without regression after each change.
Technical Specs:
- YAML Parser: Extraction rules defined in YAML — no need to touch Python to add new SQL patterns
- Stack: Python metadata-driven + React Flow UI
- Privacy: Client-side/local in import mode, or direct connect to DWH
- Scope: Object-level (column-level not supported)
- Tested with: Azure Synapse
- Deployment: Docker-ready
Result
Easy dependency analysis. More use cases to follow.
Get It
Full source released under MIT.
👉 GitRepo
Repo, live demo, and video walkthrough all linked there.
Feedback welcome — it’s yours to explore or fork.
Greetings,
Christian
Top comments (0)