dbt Discovery API turns run metadata into queryable project state

Every dbt project run generates and stores metadata on models, sources, other nodes, and execution results. The Discovery API queries that store so you can inspect the DAG and what it produces. The docs position uses such as monitoring and alerting, lineage exploration, and automated reporting, aimed at discovery, data quality, and pipeline operations.

Access paths include ad hoc queries, custom applications, and partner tools in BI, catalog and governance, and quality and observability, plus dbt features like model timing and data health tiles. You can query at environment scope for latest state or historical model runs, and at job scope for a specific run and resource type such as models or tests.

For data and AI teams wiring catalogs or agents to transformation systems, this API is a machine-readable view of production dbt meaning—nodes, results, and history—rather than only SQL in repos. Shared metadata of this kind is what governed semantic layers and interoperability tools need when they align schemas, lineage, and operational truth across systems. Prerequisites named in the extract include a multi- or single-tenant account, Starter or Enterprise plans, and dbt 1.0 or later.