Backfilled 2026-05-23. Nightly PM did not fire this night (machine asleep / daemon did not launch). Reconstructed from file mtimes + notify.log + supplemental /trading/dev/ams scan.
_(reconstructed; see the live 5/23 report for the current Top 3.)_
risk-on · breadth +0) — ~2h later than the usual ~05:41 slot. Mid-Day emailed 11:10 ET (yellow lights).config.json edit via cockpit admin UI 08:51 ET (timestamped backup written). HUD-off-limits respected.index.html + manifest.json rebuilt after the morning/midday publishes.error_log entries appeared across _auth, cockpit, _api/cockpit, mgp, app, and scorecards in both dev and stage AMS trees. No new source files — this was a testing/debugging session against the AMS properties built 5/20–5/21, with runtime errors being logged. Worth a look at those logs before the next AMS push./mgp/2026-05-22-recap.html was generated and notify.log has no recap entry for 5/22 — only morning + mid-day went out. Combined with the late morning fire, this is the second straight trading day of MGP instability, again coinciding with the Vital→independent-source migration. Action: fix the recap cron + late-fire timing before the Vital cutover; the "3 clean trading days" gate is not being met./z/public_html/mgp/2026-05-22.html + -midday.html (NO recap) + index.html + manifest.json/z/gt_dev/db_reports/smb-reports-v2/accountability/config.json — edited via cockpit (+ backup)/z/setup_daily_reporting/reports/notify.log — morning + midday sends logged; no recap