Skip to content

Start with Recce Cloud

Validate data changes and collaborate with your team with automation.

👉 Start with Recce Cloud

Setup steps:

  1. Git Platform Integration
  2. Data Warehouse Diffing
  3. CI/CD Automation

Fall back to manual if you unable to finish the setup.

Prerequisites

  1. Admin access for git platform
  2. Data warehouse credentials with read access 3.

Git Platform Integration

Connect your GitHub or GitLab repository to see all PRs/MRs in one place and validate changes before they hit production.

Setup Requirements

GitHub GitLab
• GitHub repository with dbt project
• Repository admin access for initial setup
• Active PRs with model changes		 • GitLab project with dbt project
• Project maintainer or owner access for initial setup
• Active merge requests with model changes

Note

You'll need administrative access to the Organization/Group you want to connect. Please ensure you have the necessary permissions for App installations (GitHub) or Providing a Personal Access Token (GitLab).

Connection

Connect your repository to track pull requests/merge requests and validate changes.

Note

Keep Connection setup note the same as before if there was one specific to this section.

Connection setup:

GitHub GitLab
1. Navigate to settings
2. Connect GitHub repository
3. Authorize Recce access
4. Select repository		 1. Navigate to settings
2. Connect GitLab by providing a Personal Access Token (see our directions here)
3. Connect a project by adding a GitLab Project or URL to a Recce Project

How to Use PR/MR Tracking

Once connected, Recce displays all open and draft PRs/MRs in your Recce Cloud project.

PR/MR Validation Workflow

  • View open/draft PRs/MRs in dashboard
  • Select PR/MR to validate
  • Upload PR/MR metadata (until CI/CD is configured)
  • Launch Recce to analyze changes

Data Warehouse Diffing

Do data diffing to see how changes affect your actual data. Configure your data warehouse connection to compare query results and catch issues before they impact production.

Setup Requirements

  • Data warehouse credentials with read access
  • Network connectivity to your warehouse

Supported Warehouses

  • Snowflake
  • Databricks
  • Others coming in future releases

Warehouse Connection

Configure connection to your data warehouse to enable query result comparisons.

Quick setup:

  1. In your project home, click the gear icon next to Warehouse Connection
  2. Create a new connection or select an existing one from the dropdown
  3. Your connection is linked and ready to use

For detailed connection settings, see Connect to Warehouse. Connection credentials are encrypted and secure, see our security practices.

CI/CD Automation

Set up CI/CD to automatically upload metadata and run validation checks on every PR.

Note

Available with Team plan (free trial included).

Setup Requirements

See the CI/CD sections for complete setup guides:

Automation Benefits

  • Automatic metadata upload on every PR
  • Consistent validation across all PRs
  • Reduced manual setup steps
  • Integrated PR status checks
  • Validation results directly in PR

Manual Upload Metadata

Recce shows what changed between base and current environments and helps assess potential impact. The most common case is comparing your development branch against your production or main branch to see what your changes will impact.

If you don’t use the GitHub/GitLab or havn’t setup CI/CD yet, you can manual upload

  • Web interface: Click "Update" on the session you want to update in Recce Cloud.
  • Click "Update" in base session to upload baseline metadata
  • Click "Update" in current session to upload comparison metadata
  • Click "Launch" to compare current against base
  • CLI command:
recce upload-session --session-id <your_session_id>

Find your session ID in Recce Cloud web interface when clicking "Update" on any session.

Required Files

Recce needs manifest.json and catalog.json from both base and current environments for comparison.

Base Metadata

Production environment is commonly used as the baseline, but any environment can serve as the base.

Choose one method:

Method 1: Generate locally

dbt docs generate --target-path target-base --target <your_prod_target>

Method 2: dbt Cloud
Deploy → Jobs → Production job → Recent run → Download artifacts

Method 3: dbt Docs server
Download the artifacts directly from dbt docs server:

  • <dbt_docs_url>/manifest.json
  • <dbt_docs_url>/catalog.json

Current Metadata

Use development environment or PR branch as current to compare against the base.

Choose one method:

Method 1: Generate locally

dbt docs generate --target <your_dev_target>

Method 2: dbt Cloud
Deploy → Jobs → CI job → Recent run → Download artifacts