Re-test Plan

Replays the original baseline scans at the new SHA. Must be green before any verdict flip.

• pnpm install --frozen-lockfile at the new SHA — no missing peer-deps warnings
• pnpm lint exit 0 (regression baseline — AUDIT-001 must be cleared)
• pnpm build exit 0 (tsc + Vite)
• ts-prune — zero false-positives in useApi.ts (post-split)
• jscpd — no new duplication clusters versus Phase-1 baseline
• madge / dependency-cruiser — no new cycles
• Bundle analyzer — largest chunk stays under 500 kB gzipped budget

Each deep-dive (DD-1..5) re-run end-to-end with two role accounts where applicable.

• DD-1 (Auth) — A-AUTH-001..015 + C-FE-005a/b/c each closed or risk-accepted
• DD-2 (RBAC) — end-to-end with two role accounts; admin + non-trader walk every protected route
• DD-3 (Trading command path) — paper + real sessions; both modes re-test A-MONEY-001..004
• DD-4 (Real-trade confirmation) — ConfirmRealTradeModal regression watch (PASSED in audit)
• DD-5 (Real-time data integrity) — C-RT-001 cursor replay + C-RT-002/003/004 banner UX

Verifies the bundle contains only the active brand. Gates the R2 verdict flip.

• ls dist/branding/ shows only the active brand's folder
• grep -r "brand-dimtech|brand-route07|Dimtech|Route07|dimtech" dist/ returns zero (case-insensitive)
• Build VITE_BRAND=dimtech and re-grep with TheGroup strings — symmetry check
• tsconfig.json::@branding-data resolves through Vite; malformed branding/active.json fails build cleanly

Full execution of the perf runbook. Pre-requisite: Block F decision from the read-out agenda.

• §5.1 — Initial load + TTI for three throttle profiles
• §5.2 — Real-time throughput (commit rate ≤ 20/s under peak)
• §5.3 — React Profiler scenarios A / B / C / D
• §5.4 — Log-viewer scroll FPS matrix (displayLimit 500 / 1K / 5K)
• §5.5 — Compressed 4-hour memory stability
• §5.6 — Lighthouse CI authenticated against marketmaker-app.111dimtech.com/ or non-prod