dbt-labs / dbt-labs/agents_schema
DBT_MODEL should carry the physical relation name for aliased dbt models
- Dominant language
- Python
- Stars
- 48
- Forks
- 4
- PR merge metrics
- No merged PRs in 30d
Description
#40 surfaced this: the dbt connector writes `node.name`, the logical model
name, to `AGENTS.DBT_MODEL`, and never `node.alias` or `node.relation_name`.
When a project sets `config(alias=...)`, the physical warehouse relation
differs from `name`, and the table `AGENTS.DBT_MODEL` points agents to does
not exist. The analyst skills resolve dbt models as `schema_name.name`, so
they hit this directly.
Proposal: add a nullable `alias` column to `DBT_MODEL` (or `relation_name`
for the full physical path) and have consumers use `coalesce(alias, name)`.
The change is additive, but it touches the SPEC.md table definition and
every destination writer, so we kept it out of #40.
Contributor guide
Research direction
Start with the DBT_MODEL table definition in SPEC.md, then trace every destination writer and the analyst skills that resolve dbt models. Compare the available name, alias, and relation_name values for aliased models. Done means the schema, writers, and consumers consistently resolve the physical relation without breaking models that have no alias.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- data-engineering, databases
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100