The project's test command

The project, not the loop, chooses how its tests run. This tool is the command the project already uses, found in the project before anything is run.

Finding it

Look where the project itself runs its tests, in this order:

1. CI: the job that runs the tests on every pull request. It is the command the project trusts.
2. The project's task file: a test script, target or task in the file the project runs its tasks from.
3. The README or the contributing guide: the command it tells contributors to run.
4. The test framework's default, read off the framework the project depends on.

When these disagree, CI wins: it decides whether a change is merged. When none of them names a command, ask a person rather than guess.

Running all of it, or one test

• The whole suite: the command as found. The work counts as done only when this passes.
• One test, to see a single failure quickly: every framework has its own way, for example pytest path::name, npx jest -t name or go test -run Name ./pkg. Read the framework's help rather than guess the flag.