Deep QA for Feature 2. Every AiTM detection path: visible-text clones, OCR text-as-image, reverse-proxy kits, off-domain credential relay, and negative controls.
PASS 0 · FAIL 0 / 0
Extension ON. Click a case (opens in a new tab) and watch the BDR service-worker console + dashboard.
HIGH cases redirect to the block page — that's the detection. Image cases take a few seconds (OCR).
Expected e.g. [BDR] DETECTION [high] sso-clone: Visible microsoft login on a non-microsoft domain ….
Automated mirror: node bdr/tests/sso-clone.test.mjs + phishing.test.mjs.
⚠️ You're not on localhost, so cases share one host and the arbiter dedups repeats for ~60s.
Test one case, wait ~60s (or reload after a minute) before the next, or run this page on the Mac at
http://localhost:4005/phishing/aitm/aitm-tests.html where each case gets its own
*.localhost domain and fires independently.