dbt-ci

dbt CI

GitHub Actions Steps

Full Build (CD / Production Baseline)

Use for production baseline workflows where you build everything:

If uv:

- name: Run dbt
  run: |
    uv run dbt deps
    uv run dbt build --target {TARGET}
    uv run dbt docs generate --target {TARGET}

If pip:

- name: Run dbt
  run: |
    source .venv/bin/activate
    dbt deps
    dbt build --target {TARGET}
    dbt docs generate --target {TARGET}

Incremental Build (CI / Pull Requests)

Use --state for state-based selection when comparing against a baseline:

If uv:

- name: Run dbt (modified models only)
  run: |
    uv run dbt deps
    uv run dbt build --target {TARGET} --select state:modified+ --state {STATE_PATH}
    uv run dbt docs generate --target {TARGET} --select state:modified+ --state {STATE_PATH}

If pip:

- name: Run dbt (modified models only)
  run: |
    source .venv/bin/activate
    dbt deps
    dbt build --target {TARGET} --select state:modified+ --state {STATE_PATH}
    dbt docs generate --target {TARGET} --select state:modified+ --state {STATE_PATH}

Variables

Variable	Description	Example
{TARGET}	dbt target/profile name	prod, ci, dev
{STATE_PATH}	Path to baseline manifest for comparison	target-base

State-Based Selection

• state:modified - Only models with code changes
• state:modified+ - Modified models + downstream dependencies
• +state:modified+ - Upstream + modified + downstream

Requires --state <path> pointing to a directory with baseline manifest.json.

Best Practices

• Use state:modified+ in CI to only rebuild what changed
• Always run dbt docs generate to create artifacts for comparison tools
• Match --target to your profiles.yml environment (prod, ci, dev)
• Ensure baseline manifest exists before using --state flag

Related Skills

• Use python-uv-ci skill for Python/uv environment setup steps
• Use python-pip-ci skill for Python/pip environment setup steps
• Use recce-cloud-ci skill for Recce Cloud upload/download steps