# Focused review demo · behavior and regression checks

Run date: 2026-10-05  
Command: `node --test tests/workflow.2026-10-04.test.mjs`  
Result: 11 tests passed, 0 failed.

The suite covers four-case initialization and selection, deterministic rewrite, exact-text approval, approval invalidation after edits, hard-blocked approval before and after rewrite, rejection, escalation, synthetic export without editable draft text, reset, unknown case IDs, and a regression check for the blocked-warning visibility state.

The warning regression test verifies that RC-101 is eligible and RC-102 remains hard-blocked, that the app maps the case flag to the element’s `hidden` property, and that `.blocked-note[hidden] { display: none; }` follows the grid display rule in the stylesheet. This is a source-level test; the separate October 5 browser QA confirmed the rendered states.

Browser QA confirmed RC-101 shows no blocked warning and has approval available; approving then editing cleared the approval and returned it to pending review. RC-102 showed its warning and kept approval disabled before and after deterministic rewrite. Reset restored the original cases and a single reset history entry. The desktop queue/detail layout fit without horizontal clipping; Chrome responsive emulation at 400 × 818 CSS pixels showed no horizontal clipping. The review was browser QA, not a screen-reader review or formal accessibility certification.

The standalone demo and the copy served by the website preview both pass the same 11-test suite. These checks do not establish screen-reader conformance, formal accessibility certification, security readiness, or production readiness. All cases are fictional; state is in memory only, with no network calls, customer system, or AI provider.