Triggering runs

There are two ways a suite gets run: manually, from its project page, or automatically, from a push to GitHub.

Manual runs

From a project's detail page, each detected suite has a Run button. This triggers a run immediately against the project's current environment build — if the environment hasn't been built yet (or needs rebuilding), the run waits for that to finish first.

Push-triggered runs

If your project is connected via GitHub (see Connecting GitHub), a push to its configured default branch automatically:

1. Forces a rebuild of the project's environment image.
2. Runs every suite in the project once the rebuild finishes.
3. Reports the result back to GitHub as a commit status on that push's commit.

A push to any other branch — a feature branch, a pull request branch — doesn't trigger anything. Only the default branch does.

What a run actually does

Each run gets its own container, started from the project's baked environment image. Where the framework's test runner produces a structured report (JUnit XML, Robot Framework's output.xml), Kiwami parses it back into a per-test pass/fail breakdown once the container finishes — see Reading results.

Next: Reading results.