@signalpilot-labs/dbt-versioning
ALoad when task involves dbt model versioning, creating v2 of a model, or backward-compatible model changes. Covers versions YAML config, defined_in, latest_version, and ref() with version pins.
Install
agr install @signalpilot-labs/dbt-versioning --target claudeWrites 1 file into .claude/skills/, pinned to git-545022af.
- .claude/skills/dbt-versioning/SKILL.md
Document
name: dbt-versioning description: "Load when task involves dbt model versioning, creating v2 of a model, or backward-compatible model changes. Covers versions YAML config, defined_in, latest_version, and ref() with version pins." disable-model-invocation: false allowed-tools: Bash(dbt *)
dbt Model Versioning
1. Versioned Model vs Standalone Copy
When a task says "create a v2," "add a new version," or "backward-compatible change," use dbt's native versions: config (dbt-core 1.6+). Do NOT create a standalone model with a _v2 suffix - that produces a separate node (model.project.dim_customers_v2) instead of a version node (model.project.dim_customers.v2).
2. YAML Structure
Declare versions inside the model's .yml file under the model entry:
models:
- name: dim_customers
latest_version: 1
versions:
- v: 1
- v: 2
defined_in: dim_customers_v2
Rules:
versions:is a list ofv:entries inside themodels:block, under the model name.defined_in:maps a version to its SQL filename (without.sql). Omitdefined_inwhen the filename matches the default<model_name>_v<N>.sql.latest_version:controls which versionref('dim_customers')resolves to. If the task says "not yet the primary version," setlatest_versionto the old version number.
3. File Naming
Version SQL files use the pattern <model_name>_v<N>.sql.
dim_customers.sql- original model, becomes v1 implicitly.dim_customers_v2.sql- new version.
Do NOT rename or move the original SQL file. It remains v1 without changes.
4. How ref() Works with Versions
ref('dim_customers')resolves to whichever versionlatest_versionspecifies.ref('dim_customers', v=2)pins to version 2 regardless oflatest_version.
Downstream models that need the new version must use ref('dim_customers', v=2). Models using bare ref('dim_customers') continue to get the old version until latest_version is updated.
5. Writing the v2 SQL
Read the original model's SQL first. If the v2 only renames or adds columns, write a thin wrapper:
select
customer_id,
customer_name,
account_status as customer_status -- renamed column
from {{ ref('dim_customers', v=1) }}
Do NOT duplicate all upstream logic into the v2 file when a SELECT-with-renames suffices.
6. Verification
After creating the versioned model, run:
dbt ls --select dim_customers --output json
Confirm the output contains nodes model.<project>.dim_customers.v1 and model.<project>.dim_customers.v2. If these nodes are missing, the versions: YAML is wrong.
Rules
- NEVER create a standalone
_v2model withoutversions:YAML - dbt will not register it as a version. - NEVER rename or move the original SQL file when adding a new version.
- ALWAYS set
latest_versionexplicitly - omitting it defaults to the highest version number, which may break downstream refs. - ALWAYS read test files in
tests/before writing models - they reveal required dbt features the task may only hint at.
Trustgrade A
- passBody integrity
Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.
- passType matchnot applicable to this artifact type
Whether the artifact is really the kind of thing its metadata claims it is.
- passFreshness
How long since the source repository was last pushed to.
- passPrompt injection
Scans the artifact's own text for instructions aimed at your agent rather than at you.
- passLicense
Whether the source repository declares an SPDX license permissive enough to redistribute.
How the grade is calculated
Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:
- Aevery check passed
- Bone warning
- Ctwo warnings
- Dprompt injection or body integrity failed, or three warnings
- Fone of those failed, and something else is wrong
These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.
Versions
git-545022afeea52026-07-31