The nightly PM went dark 5/20–5/22 (3 consecutive missed fires;nightly-startup.logjumps from 5/19 DONE straight to tonight). All three days were backfilled in this run. AMS-tree work was recovered via a supplemental/trading/dev/amsscan — the SKILL's default scan roots don't cover it, so the 5/20–5/21 AMS deliverables would otherwise have been invisible.
mgp-drop-vital-independent-source). The plan was "verify 3+ clean trading days, then cancel Vital" — instead the last two trading days both broke. Fix the cron/notify path + restore the recap edition first; do not cancel Vital or flip cron fully to –source=independent until 3 genuinely clean days are logged.update_freshness.py written, just needs cron + DB-source half. It is exactly the surface that would have flagged the 5/20–5/22 nightly-PM gap and the MGP delivery misses in near-real-time instead of three days late. This week's regressions are the argument for finally shipping it.playbooks/ + system/ directories appeared (23:12–23:13 ET) in both dev /trading/dev/ams/ and stage /trading/ams/ — the likely start of the next two AMS property ports (Playbooks + System hub) following Cockpit (5/19), Scorecards (5/20), and MGP (5/21). Empty scaffold so far; no content files yet.james-monthly-reports/may-2026/, /vwap-bands/, /vwap-bands/param-surface-db/, top-level index) — auto-pipeline, clean.p0-freshness-dashboard — ~8 days deferred. See TODAY #3.sync-from-legacy.php only; entitlement/scope-aware generation pending._auth/cockpit/mgp/scorecards threw during testing; review before the next AMS push./z/trading/dev/ams/playbooks/ + /system/ — new dirs (23:12) → mirrored to /z/trading/ams//z/public_html/scorecards/trade_scorecard_2026-05-23.html + manifest/z/gt_dev/db_reports/james-monthly-reports/ — may-2026 + vwap-bands + param-surface-db + top index regen