Capabilities of dbt that the library does not have yet
mediumopenHeld: models with order and lineage, view/table/incremental/ephemeral (0023), generic and singular data tests with severity, thresholds, where, limit, failCalc, stored failures, tags and selection (0052), docs generation (documentation extension). Missing, in the order of their value: sources with freshness checks; seeds (typed CSV loads, with column types and quoting); snapshots (type-2 history by `check` or `timestamp` strategy; the history extension is not this); unit tests of models with given rows and expected rows; incremental strategies beyond append (merge, delete+insert, insert_overwrite, microbatch); model contracts and constraints checked at build; exposures, groups, access and versions of models; run hooks (pre/post, on-run-start/end); analyses; state-based selection and defer (`state:modified+`); a macro and package mechanism (here: TypeScript modules, so a package is an npm package of generic tests and models, which needs a documented convention); a run results artifact (`run_results.json`, `manifest.json`) and the semantic layer. Each is to be taken as a spec of its own.
| Library | Data |
| Severity | medium |
| Priority | P2 |
| Status | open |
| Size | xl |
| Goals | G1, G2 |
| Raised | 2026-11-19 |