What next?

Your first test runs. Rerunning it is one key or one command, and that gets addictive quickly. Here is where the rest of CAT is — the How-to guides, in the order most projects need them.

Decide what to test #

The dumb test first — is there data where data should be — then the rules the data must obey, then the comparison between two systems that should hold the same data. The Test patterns guides walk that order: smoke tests, finding problems with set is empty, comparing data across systems, incremental loads, schemas, compliance rules. When you are not sure which expectation a test needs, Which expectation when decides it.

Keep the project in order #

One file is enough for a start and not for long. Organize and run tests shows how to split a project into several files, keep test definitions in a database table or an Excel sheet the business maintains, keep passwords out of the file, run only a suite or a tag, and let CAT generate a test per table from a metadata query instead of writing each one.

Test your data platform #

Power BI semantic models and measures, Databricks, Microsoft Fabric, Dataverse, CSV and Excel files — each has its own way to connect and its own tests that pay off. Data platforms has a guide per platform.

Run it unattended #

A scheduled run is one command in a pipeline step. Pipelines and schedulers has the step for Azure DevOps, GitHub Actions, GitLab, Jenkins, cron, Windows Task Scheduler and SQL Server Agent, the plan it needs and where the secrets live.

Keep the results #

Without an output a run leaves nothing behind. Outputs write the results to Excel, JSON, YAML, JUnit or TRX files, or one row per test into a database table — and that table is what Track test results over time turns into a pass-rate trend, a dashboard and an alert.

Let an AI agent write the tests #

CAT is built to be used from a coding agent: the CAT CLI runs the project, the CAT skills tell the agent how to write it. CAT and AI agents explains the loop and how to install the skills into Claude Code, Cursor, Codex or Copilot CLI.

Look things up #

Every key of the project file, every expectation, every data source provider and every output is in the Reference, with a section per tool — CAT Studio, CAT CLI, the PowerShell module, the Python module.