lineage-analyzer

Lineage analysis traces upstream sources and downstream consumers in dbt projects by parsing {{ ref }} and {{ source }} references from SQL files or manifest.json. The analyzer builds a bidirectional dependency graph to reveal complete data lineage, showing which models feed into others and where data originates.

Yes. The analyzer identifies circular dependencies—cases where models reference each other in cycles—by applying cycle detection to the constructed dependency graph. This catches configuration errors that would otherwise fail silently during dbt runs.

Lineage output is available in three formats: tree view for hierarchical structure, detailed format for complete relationship information, and Mermaid diagrams for visual graph representation. Choose the format that best fits your impact analysis or debugging workflow.

By mapping upstream and downstream relationships, impact analysis shows which models depend on a changed source or intermediate model. This reveals the blast radius of schema or logic changes before deployment, reducing debugging time and unintended downstream breakage.

No. The analyzer scans the models/ directory and parses {{ ref }} and {{ source }} references directly from SQL files. manifest.json is optional and used when available to enhance dependency resolution, but the core lineage trace works without it.

Mermaid diagram output scales well for visual inspection of dependency graphs, showing the complete flow from sources through transformations to fact tables. Tree and detailed formats work better for programmatic analysis or drilling into specific lineage paths.