Test

Write test specs for your capabilities, group them, and run them with pass/fail reports in a panel at the bottom of the editor.

What you can do

• Group your test specs by the capability they check, with per-group stats and a "Run all" button for each group.
• Run everything at once, run one group, or run a single spec — then reset results when you want a clean slate.
• Read a summary across all specs: total specs, total cases, passed, failed, skipped, and pass rate.
• Expand any spec to see its last report or a preview of the cases that haven't run yet.
• Score outputs with built-in checks: equality, contains, pattern match, numeric ranges, type, property, and array length.
• Run your project's own test suite next to them. Install a test engine such as Python Tests (pytest) and it appears here as its own row, with the failing tests listed and the counts folded into the header.

Test engines

This panel carries no engine of its own. An engine is a separate extension that knows how to run one kind of test suite. Install one and it shows up here; the same run can also be a gate in an MLOps pipeline, so a red suite stops the pipeline and both places report it identically.

Requirements

No workspace or kernel needed. Everything lives in memory and resets when you reload.

Getting started

Open the Test panel from the activity bar. Install a test engine such as Python Tests before you can run anything.