Roadmap and FAQ
Roadmap
| Step | Item | State |
|---|---|---|
| M1 | SQLGlot bridge (isolated, pinned, optional) and dialect fixtures | Planned |
| M2 | MERGE, UPDATE, PIVOT, table functions | Planned |
| M3 | OpenLineage import, dbt version validation, compiled SQL | Planned |
| M4 | Git-aware schema diff, SARIF | Planned |
| M5 | MCP Inspector validation, native IDE checks | Planned |
| M6 | Signed evidence, SBOM, Scorecard | Planned |
| M7 | Hosted dashboard with auth and RBAC | Not started |
FAQ
Does it connect to my database? No, never.
Does it send my SQL anywhere? No network calls exist in the core.
Is the lineage guaranteed correct? No. It is evidence for review inside a documented subset; unknowns are listed.
Why is something "possible" rather than "confirmed"? The column is used in a filter, join or group, or only a table-level dependency was found.
Can it read dbt projects? Manifest metadata, yes. Compiled SQL scanning works on compiled files. Jinja inside SQL is flagged.
Is the audit independent? It is a self-audit. It prints its limits and how to reproduce it.
What is the 100% gold score? Author-labelled regression corpus of 54 cases. Not a claim about your SQL; run audit --root.
Does an AI agent run the tool for me? Skills instruct a host agent; whether your host runs them natively is untested.
Package name? lineagesutra-ai; commands lineagesutra and lineagesutra-ai.