DataCI

CI/CD for analytics engineering.

DataCI posts a detailed report on every dbt pull request — showing impact analysis, test coverage, and risk assessment — so reviewers know exactly what changed and what it affects.

No database connection needed. No profiles.yml setup. Just point it at your dbt project.

---

What You Get

Every PR that touches dbt models gets a comment like this:

Changed Models

Model	Change
stg_orders	modified

Impact Analysis

4 downstream model(s) affected | Risk: MEDIUM

• int_order_items
• fct_orders
• dim_customers
• fct_revenue

Test Coverage

57.1% (4/7 models tested)

Changed models missing tests: stg_payments

---

Quick Start

Add this workflow to your dbt project — just 2 steps:

# .github/workflows/dataci.yml

name: DataCI
on:
  pull_request:
    paths:
      - 'models/**'
      - 'macros/**'
      - 'tests/**'
      - 'dbt_project.yml'

jobs:
  dataci:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - name: Run DataCI
        uses: tripleaceme/dataci@v1
        with:
          github-token: ${{ secrets.GITHUB_TOKEN }}

That's it. No dbt compile, no profiles.yml, no adapter install. DataCI handles manifest generation automatically.

---

Inputs

Input	Description	Required	Default
github-token	GitHub token for posting PR comments	Yes	${{ github.token }}
project-dir	Path to dbt project root	No	.
manifest-path	Path to a pre-built manifest.json (skip auto-generation)	No	(auto-generated)
dbt-version	dbt-core version for manifest generation	No	1.9.0
fail-on-missing-tests	Fail if changed models have no tests	No	false
coverage-threshold	Minimum test coverage % (fails if below)	No	0

---

Usage Options

Zero-config (recommended)

DataCI auto-generates the manifest — no dbt install or profiles.yml needed:

- uses: tripleaceme/dataci@v1
  with:
    github-token: ${{ secrets.GITHUB_TOKEN }}

With pre-built manifest

If your CI pipeline already runs dbt compile, pass the manifest directly:

- uses: tripleaceme/dataci@v1
  with:
    github-token: ${{ secrets.GITHUB_TOKEN }}
    manifest-path: 'target/manifest.json'

With quality gates

Block PRs that don't meet your standards:

- uses: tripleaceme/dataci@v1
  with:
    github-token: ${{ secrets.GITHUB_TOKEN }}
    fail-on-missing-tests: 'true'
    coverage-threshold: '60'

---

Features

Impact Analysis

When you change a model, DataCI traces the full downstream dependency graph and tells you:

• How many models are affected
• Which specific models are downstream
• Risk level (LOW / MEDIUM / HIGH) based on blast radius

Test Coverage

DataCI calculates what percentage of your models have at least one dbt test and flags:

• Overall project coverage
• Changed models that are missing tests

Quality Gates

Set fail-on-missing-tests: true to block PRs where new/changed models don't have tests. Set coverage-threshold to enforce a minimum coverage percentage.

Smart Comment Updates

DataCI posts one comment per PR and updates it on subsequent pushes — no comment spam.

Zero-Config Manifest Generation

DataCI reads your dbt_project.yml, creates a temporary dummy profile, and runs dbt parse to generate the manifest. No database connection or credentials needed — dbt parse only reads your SQL and YAML files.

---

How It Works

1. DataCI reads dbt_project.yml and auto-generates a manifest (or uses one you provide)
2. It parses the manifest to build the full dependency graph
3. It diffs against the base branch to find changed models
4. It traces downstream impact and calculates test coverage
5. It posts (or updates) a single Markdown comment on the PR

---

Requirements

• A dbt project with dbt_project.yml
• GitHub Actions enabled on your repository

That's it. DataCI handles the rest.

---

Roadmap

• Impact analysis
• Test coverage reporting
• Quality gates (fail on missing tests / low coverage)
• Zero-config manifest generation
• AI-powered SQL review (Claude)
• Query cost estimation (BigQuery / Snowflake)
• Lineage visualization (Mermaid diagrams)

---

License

MIT