Add a dbt_unit_test_gen tool and dbt-unit-tests skill that generates dbt unit tests from manifest + compiled SQL.
Problem
dbt unit testing (shipped v1.8) has near-zero adoption because writing test YAML is tedious — you must identify every upstream ref, mock each with correct column types, and manually compute expected outputs.
Solution
A tool that:
- Parses manifest for model info, deps, columns, descriptions
- Uses column lineage to understand input→output mappings
- Detects testable SQL constructs (CASE/WHEN, JOINs, NULLs, division, aggregation, incremental)
- Generates type-correct mock data with scenario coverage (happy path, null edge, boundary, incremental)
- Assembles complete YAML via the
yaml library
- Returns rich semantic context (descriptions, lineage) for the LLM skill to refine values
Reuses existing infrastructure: parseManifest(), dbtLineage(), schema.inspect, sql.optimize, manifest caching.
Add a
dbt_unit_test_gentool anddbt-unit-testsskill that generates dbt unit tests from manifest + compiled SQL.Problem
dbt unit testing (shipped v1.8) has near-zero adoption because writing test YAML is tedious — you must identify every upstream ref, mock each with correct column types, and manually compute expected outputs.
Solution
A tool that:
yamllibraryReuses existing infrastructure:
parseManifest(),dbtLineage(),schema.inspect,sql.optimize, manifest caching.