authoring-e2e-test-suite

Derive one executable scenario per journey with explicit setup, ordered steps, and assertions tied to acceptance criteria. Each scenario records steps_total and steps_failed, then runs against the target environment through the CI/test runner with a browser driver.

Re-run suspect scenarios a bounded number of times and mark any scenario with differing results as flaky. Compute flaky_rate as the fraction of flaky scenarios, exclude them from a PASS, and report them for stabilization since a flaky green is not a pass.

PASS requires zero journey or assertion failures and no flaky scenarios, all backed by real runner evidence. FAIL applies to any failed or flaky scenario or masked result, since uncertain is treated as a fail, and ERROR means the suite could not execute.

No, it only derives scenarios, executes them, and reports results. Failures route to a fixer such as progressive-bug-hunter, and the verdict feeds a named human verification layer rather than auto-approving.

Do not use it to design the QA plan, run backend or frontend unit tests, or run integration tests, since dedicated skills cover those levels. It also requires a reachable running target environment, so it cannot operate without one.