Practical notes on browser testing, coding agents, failed tests and test maintenance.
Functional verification and security review answer different questions about an AI-written change.
Read post →A replacement target needs evidence that the intended action and outcome stayed the same.
Read post →A useful follow-up names the failure scenario, fixture and result the test must verify.
Read post →Role, label and context help describe the intended control; an outcome assertion still does the real verification.
Read post →Review a test’s purpose, failures and ownership before deleting it or keeping it by habit.
Read post →The agent can repeat a flow; people still decide which outcome matters.
Read post →Keep the test, environment and result together; compliance conclusions require a separate review.
Read post →A framework migration can improve a workflow, but shared data and weak assertions can follow you.
Read post →A moved control, stale fixture and changed business rule need different fixes.
Read post →A faster implementation workflow needs a repeatable way to verify the customer journey.
Read post →Compare the environment, fixtures and observed state before blaming the browser test.
Read post →Ask to see the failed test, the unchanged assertion and the evidence behind any healing.
Read post →Measure the work around a failure: investigation, reruns and the cost of ignoring the gate.
Read post →