Concepts

Project                    an application, such as "Customer Portal"
  ├── Test definition      which image to run, with which command, timeout, and limits
  ├── Environment          where to test: variables, secrets, a concurrency limit
  └── Schedule             when: a test definition × an environment × a cron expression

Run                        one execution of a test definition against an environment

Project

A project groups everything that belongs to one application. Projects, test definitions, and environments each have a slug (customer-portal, e2e, staging), which appears in URLs and is how the API names them.

Test definition

A test definition describes what to run: the suite’s image (ghcr.io/acme/portal-e2e:1.4.2), an optional command, a timeout, and resource limits. A project can have several, for example a fast smoke suite and a full regression suite.

See Test definitions.

Environment

An environment describes where to test: a target system such as staging or production, given to the suite as environment variables (BASE_URL, a test user’s password). Values are encrypted at rest; secret ones are also hidden in the UI and masked in logs.

Environments belong to a project, so “production” of the Customer Portal and “production” of the Billing app are different environments, with their own variables and their own concurrency limit.

See Environments and secrets.

Schedule

A schedule runs one test definition against one environment on a cron expression, in a time zone of your choice: “the E2E suite against production, every day at 06:00 Europe/Vienna”.

See Schedules.

Run

A run is one execution. It is created by Run now in the UI, by a schedule, or through the API, and all three go through the same pipeline: the run is queued, admitted when the concurrency limits allow, executed in its own container, and finished with a status.

A run records exactly what it executed: the image as configured, the digest it resolved to, and the command. Changing the test definition afterwards does not change past runs.

Statuses

Status	Meaning
queued	Created, waiting to be admitted under the concurrency limits
preparing	Pulling the image and creating the container
running	The suite is executing
passed	The suite completed and all tests passed
failed	The suite ran, and one or more tests failed
timeout	The suite ran longer than the test definition’s timeout
cancelled	A user (or the API) cancelled it
error	TestFleet could not run the suite: registry, image, Docker, a crash outside the tests, or the memory limit

The difference between failed and error matters: failed says something about your application, error says something about the infrastructure. Notifications, dashboards, and CI scripts treat them differently. See how the final status is decided.

Users and roles

Everyone logs in. Members can do everything around testing: projects, test definitions, environments, schedules, and runs. Admins also manage users, registry credentials, and notification channels. See Users and access.