Stand up the deep-history minute-bar adapter (Polygon/Databento). The v2.4 matrix verdict just confirmed this is the blocker, not a nice-to-have: every filter still loses on the one choppy ~14-day window kti can serve, so no v2.4 verdict is trustworthy until 6–12 months × the full basket runs. The engine already abstracts bar_source — drop-in adapter, and it now gates the whole J3P arc.
Sprint 2 hygiene — advance the pointer, roll the carryovers, file the planned items. The 7AM sprint brief flags active_sprint_id still reads Sprint 1 (window closed Fri 6/26), and the 6 planned Sprint-2 items live only in sprint-plan.html, not in todos.json. Until they're filed and the 8 carryover P0/P1s roll forward, the cockpit can't track them and planning has nothing to commit against. Pure tooling.
Live smoke-test J3P v2.2 (needs James). Backtests run on simulated plays; real edge accumulation can't begin until one live J3P alert lands a row in alerts_tradingview.indicator_alerts. Attach the "Any alert() function call" alert with the webhook secret. Still the gate for the whole indicator→g8trade arc.