E2E Tester agent

The E2E Tester agent thinks in user journeys. It defines end-to-end scenarios that cover the flow a user actually sees, from the first click to the final result, and reports where coverage is missing.

Where it runs

In Feature SDLC it runs after the Backend Engineer and Frontend Engineer have described their changes, so it can test the joined-up feature. In Release Readiness it is one of four checks before the Architect's go/no-go, alongside the Reviewer, the Security Auditor and the Dependency Auditor.

Inputs

• The session task.
• A read-only snapshot of relevant files from the project's repository, including any existing end-to-end tests.
• The earlier steps of the session: requirements and acceptance criteria in Feature SDLC, or the correctness and security reviews in Release Readiness.

Outputs

A written set of scenarios, each with:

• The starting state and the steps a user takes.
• What the user should see at each point.
• Failure paths, such as a network error or an expired session.

It also lists gaps: important flows with no end-to-end coverage today. The code pass that writes the change reads its scenarios and may add test files for them to the diff. The verify step runs the project's own test command; whether that includes end-to-end tests depends on how your project is set up.

Example run

Illustrative example, not a recorded session. In Release Readiness before a billing release, the E2E Tester lists five journeys that must still work: sign up and start a trial, upgrade with a card, upgrade with a card that is declined, downgrade at the end of a period, and view past invoices. It finds existing tests for the first two only, and flags the declined-card path as the highest-risk gap because the release changes the payment error handling. The Architect's go/no-go weighs that gap.

Limits

• It does not run a browser, click through your app or record videos.
• Its view of your test suite is the files it is shown. It can report a gap that is covered somewhere it did not see.
• Whether new end-to-end tests run during verification depends on your project's test command and environment. See quality gates.