Morning Brief · ATLAS Standup

Monday 2026-07-27 ET · next standup Tue 2026-07-28 · 8:30 AM ET · published 2026-07-27 5:47 PM ET · all times America/New_York (ET)
⚠ This brief fired ~10 hours late. Scheduled slot 7:53 AM ET · actually ran 2026-07-27 5:47 PM ET. Every Claude scheduled task on this machine burst-fired together between 5:35 and 5:47 PM ET today — the machine was asleep through the whole morning window.
Same pattern as 7/13, 7/14, 7/18 and 7/22. The nightly PM at 11:04 PM ET has now missed 7 consecutive nights (7/20–7/26): no /todos/daily/ report was written for any of them, and todos.json has not been touched since 2026-07-20 1:06 AM ET. The ledger below is 7 days stale. Read it as “last known state”, not “current state”.
0
shipped overnight
29
stale blockers
36
open P0
71
open P1
171
open total
7d
ledger frozen
6
freshness alerts

What shipped overnight completed_at = 2026-07-26 (Sun) ET

0
No task carries completed_at = 2026-07-26.
The last completion recorded anywhere in the ledger is 2026-07-19 — 8 days ago.
07-06
1
07-07
1
07-08
1
07-09
1
07-10
3
07-11
3
07-12
5
07-13
5
07-14
2
07-15
1
07-16
2
07-17
1
07-18
1
07-19
5
07-20
0
07-21
0
07-22
0
07-23
0
07-24
0
07-25
0
07-26
0
Read this honestly: the seven red zeros are not proof that nothing was built. They are proof that nothing was recorded. The nightly PM is the only writer to this ledger and it has been dark since 7/19. Real work almost certainly shipped in that window (the a2 dev trees and the J3P/MGP surfaces kept running) — it is simply invisible to every Atlas surface. Recovering that week is a standup item in its own right.

Stale blockers owner ≠ claude · pending · P0/P1 · age ≥ 7 days · 29 items

AgeProjectPriOwnerTitle
101dAtlasP0user6/6 indicator_alerts DDL audit: dump schema, commit as ground truth
psql -c '\d alerts_tradingview.indicator_alerts' > /z/trading/a1p_journal/docs/indicator_alerts_schema.txt. Needed before any 'add missing webhook fields' work.
100dMultiTraderP0userUSER: 3 steps to finish journal persistence
1) Run add_scorecard_columns.sql 2) Confirm user_id=2 and db=trsports_tradingjournal_dev match your server (now pre-set in save_journal.php from gen_scorecard.py) 3) Regenerate a scorecard…
100dMultiTraderP1userRe-run gen_scorecard.py ON a2 (or via tunnel) and confirm enrichment populates
Expected: 'DB enrichment: N entries with regime/playbook tags' in log, non-zero enriched count on scorecard.
97dAtlasP1userDeploy z_activity_scanner.py on each Windows machine via Task Scheduler
Script at Z:\public_htmlctivity\z_activity_scanner.py. Import task_scheduler_import.xml on each PC. Runs every 15min, writes COMPUTERNAME-HHMM.json to /activity/YYYY-MM-DD/. User must deplo…
96dMultiTraderP0userUSER: run scientific_layer migrations 001–009 via psql tunnel
Run in order: set_db.sh start, then psql -f 001...009 against trsports_tradingjournal_dev. Unlocks: p0-daily-summary-view (3/6), sizing calculator DB wiring, rolling_stats, setup_performance…
82dAtlasP1userCron /system/update_freshness.py to keep freshness strip auto-refreshing
Script ships ready. Suggested cron: */30 9-17 * * 1-5 python /z/public_html/system/update_freshness.py --root /z/public_html (every 30 min during trading hours). Closes the silent-regression…
82dMultiTraderP1userTrader-of-the-Week scorecard review session (rotating: Nolan / Preston / Brian / Thomas)
Weekly cadence. Pick a trade day, review it with one of the 4 internal traders (rotate through: Nolan -> Preston -> Brian -> Thomas). Output: annotated /playbooks/ entry showing the trade th…
82dMultiTraderP1userFirst-scorecard feedback session with each of the 4 traders
After multi-trader-scorecards ships: schedule 4 short sessions (one per trader). Each trader sees their first scorecard, reads the journal/grading + classifier output, gives feedback on what…
82dStratLibP1alexCodify Range-Break strategy rules from Alex's Notion (entry / stop / target / best-when)
Alex sent the Range Break Notion. First step in onboarding him to Strategy Library. Translate the Notion rules into the same JSON shape used by the 16 catalog entries on /strategies/ (id, fa…
82dStratLibP1alexRange-Break: walk-forward backtest published at /backtests/range-break/
Once rules codified (alex-range-break-rules). Same walk-forward framework as fl2-band-trail and vwap-pull-in. Reuses gen_walkforward.py. Publishes: IS/OOS Spearman, deflated Sharpe, per-mont…
82dStratLibP1alexRegister Range-Break in /strategies/ Strategy Library + canonical catalog
Final step. Add range-break card to /strategies/ index.html + entry to the (about-to-be-extracted) catalog.json (see catalog-refactor-shared-json — order matters: do that refactor first so…
82dClassifierP1robertoLive Classifier HUD: auto-refresh classify/<SYMBOL>.json during market hours
James-live HUD shipped 5/5 but the classify/*.json files are static. Wire a refresh job: every 1-min during RTH, walk watchlist, pull live SD position from indicator_alerts, recompute the 6-…
82dClassifierP1userReview Roberto's Claude artifact + slot his real work into the right property
Roberto sent his Claude work at https://claude.ai/public/artifacts/1b9da8d6-6627-4f4b-b325-2c65a33bf4cc (claude.ai artifacts are 403 from WebFetch — needs a James-side pass). NOT MTF VP Sc…
81dAcct-v2P1jimboMove accountability-v2 Node services from nohup → Windows Task Scheduler (production-grade)
Per BUILT.md 'Where to go next' #1: services (position-projector, pnl-engine, fills-poller, marks-fallback) are running via nohup/PowerShell background. Add Task Scheduler entries beside the…
81dAcct-v2P1jimboWire v1 fills-poller to also POST into v2 /api/ingest.php (dual-write transition)
Per BUILT.md 'Where to go next' #2: today's fills already arrive into alerts_tradingview.fills via v1's poller; v2's position-projector picks them up via LISTEN/NOTIFY. But for double-safety…
81dAcct-v2P1userUpload AccountabilityTickBridge.py to KITE forward-test + verify ticks flow into marks_live
Per BUILT.md 'Where to go next' #3: ktg.Kite().strategy_upload() the bridge. Then watch marks_live table fill in for the dynamic symbol set returned by /api/bridge-symbols.php. James-side be…
81dAcct-v2P1userTrade Action Requests feature — finalize (close/stop intent flow: hud-orderlab + migration 021 + trade-action.php)
Per hud-orderlab survey 2026-05-07: this feature is a 3-piece set — (1) SQL migration 021_trade_action_requests.sql defines the table, (2) api/trade-action.php endpoint inserts into it (re…
81dAcct-v2P1userExecute dev→prod promotion sequence per docs/PROMOTE-DEV-TO-PROD.md
Doc PROMOTE-DEV-TO-PROD.md exists in dev/docs — implies promotion isn't a one-step copy but a sequenced procedure. Read the doc, schedule the promotion window, follow steps. Prerequisite: …
81dAcct-v2P0jimboBUG: Risk Bump Progress bar does not respect the selected date range (always shows 5-day rolling, even when range = full month)
Reported by James 2026-05-07 with two screenshots from the live DEV HUD (https://smb.a1prospect.com/live-accountability-v2-dev/hud/?t=james&range=c...). REPRO: select James + date range 2026…
80dAtlasP1userTest mklink workaround for Claude Code 'suspicious path pattern' prompt on WebDAV-backed projects (unblocks nightly PM routine + agentic sessions)
Claude Code's safety layer flags suspicious Windows path patterns and forces manual approval on EVERY file write — even with 'Allow bypass permissions mode' enabled — because the WebDAV …
80dAcct-v2P1userDecide whether Risk Manager HUD (/accountability/risk.html, mock) is redundant with Accountability HUD (/live-accountability-v2/hud/) — merge or retire
James flagged 2026-05-08 to constantly scan for duplication. The Risk Manager HUD card on /system/ (built 2026-04-21, mock data, 'awaiting live API endpoint from smb subdomain' per risk-hud-…
75dAtlasP0jamesCompliance follow-up: delete (or hide) historical MGP pages 4/17 through 5/12 — they still contain Vital Knowledge text
PENDING JAMES'S APPROVAL. Today's three pages (5/13 morning + midday + recap) are clean. But ~40 historical pages from 2026-04-17 through 2026-05-12 still on the public web at https://trspor…
75dAcct-v2P0jimboBUG: Ticket Meter denominator doesn't scale with date range (shows 1901 / 30 on 13-day MTD range)
Reported by James 2026-05-13 with screenshot from live DEV HUD. REPRO: select James + date range 2026-05-01 → 2026-05-13 (MTD). OBSERVED: TICKET METER shows '1901 / 30' with the label '100% …
73dAtlasP0JamesBUG: MGP auto-fire skipped 5/14 mid-day + recap; 5/15 morning ran ~2h45m late
notify.log shows 5/14 06:41 send then nothing until 5/15 08:35 (normal slot 05:48). 5/14 mid-day + recap files exist but are manual evening regens that did not email. Two-day degradation. In…
65dAtlasP0jamesNightly PM daemon went dark 5/20-5/22 (3 consecutive missed fires) — fix wake/launch
The accountability system's own nightly reporter did not run on 5/20 (Wed), 5/21 (Thu), or 5/22 (Fri). nightly-startup.log jumps straight from 5/19 23:21 DONE to 5/23 23:05 LAUNCHED — no LAU…
37dAMSP0userAMS: seed main.users with trader rows so identity_fn returns non-zero
Per AMS Phase 0 status doc: p0-identity-fn shipped + 5/5 smoke tests pass, but resolve_identity() returns 0 rows because no users in main.users have usernames matching trader short_names (ja…
37dAMSP0userAMS: resolve 7 open decisions D1-D7 (canonical domain · auth stack · DB strategy · sharing · risk_hud · 2FA · daily-stop)
Per AMS README §'Open decisions': D1 canonical domain (rec: smb.a1prospect.com), D2 auth stack (rec: stay PHP), D3 DB strategy (rec: build in _dev, promote together), D4 sharing default (rec…
36dPinePBP0userTradingView setup guide for James — paste Pine script, configure alerts, test
Click-by-click instructions. Save indicator to user's library. Add to chart. Configure alert with webhook URL. Test fire. ~15 min one-time.
36dPinePBP0userJames pastes finalized Pine script into TradingView + configures the 3 alerts
After Atlas delivers the script + setup guide, James spends ~15 min configuring TradingView. Each play gets its own alert (3 alerts total) so they can be toggled independently.

Open P0s by project 36 total · top 3 per project

Atlas13 open P0
  • P0 claude · 101d 1/6 Data-freshness dashboard: /system/freshness.json + banner on hub
  • P0 claude · 101d 2/6 Pipeline orchestrator: run_nightly.py calls every generator, logs to one place, fails loud
  • P0 user · 101d 6/6 indicator_alerts DDL audit: dump schema, commit as ground truth
SIP7 open P0
  • P0 claude · 37d Catalyst feed: Vital Dawn + Finnhub news + SEC 8-K + Benzinga RSS into per-symbol catalyst_score
  • P0 claude · 37d Publish ranked watchlist.json + /sip/index.html (top 15-20 · refreshes pre-market + every 30m RTH)
  • P0 claude · 37d Integrate SIP into Trigger Listener: gate alerts on watchlist · enrich payload with score + catalyst + RRG
MultiTrader4 open P0
  • P0 claude · 101d 3/6 main.daily_summary SQL view: kills duplicate computation
  • P0 user · 100d USER: 3 steps to finish journal persistence
  • P0 user · 96d USER: run scientific_layer migrations 001–009 via psql tunnel
AMS4 open P0
  • P0 claude · 101d 4/6 Identity function: resolve_user(account_id) -> (user_id, trader_name)
  • P0 user · 37d AMS: seed main.users with trader rows so identity_fn returns non-zero
  • P0 claude · 37d AMS: create main.remembered_devices table (migration 002) — Phase 1 hard dependency
PinePB4 open P0
  • P0 claude · 36d Codify VWAP Pull-In + Trend Trade + Backside Scalp rules in Pine-compatible terms
  • P0 claude · 36d Write Pine v5 indicator combining all 3 playbooks with toggle inputs
  • P0 user · 36d TradingView setup guide for James — paste Pine script, configure alerts, test
Acct-v22 open P0
  • P0 jimbo · 81d BUG: Risk Bump Progress bar does not respect the selected date range (always shows 5-day rolling, even when ra
  • P0 jimbo · 75d BUG: Ticket Meter denominator doesn't scale with date range (shows 1901 / 30 on 13-day MTD range)
StratLib1 open P0
  • P0 claude · 101d 5/6 Param Surface as strategy registry: /param-surface/ index + per-strategy grid pages
Classifier1 open P0
  • P0 claude · 50d Liberator minute_bars entitlement LOST (27->10 datasets) — Stage B mechanism backfill now BLOCKED on data

Freshness alerts every monitor that is supposed to be watching

SourceStatusAgeDetail
/system/freshness.jsonDEAD81 daysGenerator last wrote 2026-05-07 9:38 PM ET. Every value on this file is 81 days old — the two “stale” rows below are frozen artifacts, not live signal. Superseded by the App Watchdog, but nothing removed it from the hub.
  ↳ Live Classifier HUD (from freshness.json)stale54.9 h (as of 5/7)/scorecards/james-live.html — frozen reading, not verified today.
  ↳ Nightly PM Report (from freshness.json)stale37.2 h (as of 5/7)/todos/daily/ — frozen reading, but independently CONFIRMED still true today (see below).
/system/atlas-flow-full.jsonSTALE + RED4 daysLast run Thu 2026-07-23 8:30 PM ET. overall_status = RED — update_scanners FAILED, run_sip_daily skipped (11 pass / 1 fail / 1 skip).
/system/atlas-flow-status.jsonSTALE3 daysLast run Fri 2026-07-24 5:00 PM ET, 16/16 pass. Green, but 3 days old — it is not watching anything right now.
/system/app-status.htmlSTALE19 daysLast written 2026-07-08 12:01 PM ET on trsportsacademy. The 30-min watchdog that was supposed to replace freshness.json has itself stopped writing here.
/todos/todos.jsonFROZEN7 daysLast write 2026-07-20 1:06 AM ET. This is the source every Atlas surface reads — the whole PM ledger is 7 days behind.
The watchers are all asleep. There is no monitor on this system that has reported in the last 72 hours. freshness.json is 81 days dead, the App Watchdog page is 19 days dead, and both atlas-flow probes stopped 3–4 days ago — one of them RED on its last breath (update_scanners failed). This is exactly the failure class that p0-freshness-dashboard (focus #1 below, open 101 days) was filed to prevent, which is why it keeps ranking first.

Today's standup focus open P0/P1 ranked by priority × age

1
Atlas P0 1/6 Data-freshness dashboard: /system/freshness.json + banner on hub
owner claude · open 101 days · status pending · weight 95.6
STATUS 2026-05-06: schema scaffolded at /system/freshness.json AND seeded with property mtimes (9 fresh / 1 stale / 3 external). update_freshness.py written. Outstanding: cron the script + DB-source half. Bumped 6+ days; would have caught all 3 silent regressions in the last 9 days.
2
Atlas P0 2/6 Pipeline orchestrator: run_nightly.py calls every generator, logs to one place, fails loud
owner claude · open 101 days · status pending · weight 95.6
Replaces cron uncertainty. Log to /z/logs/nightly-YYYY-MM-DD.log. Any step returning empty = non-zero exit + alert.
3
MultiTrader P0 3/6 main.daily_summary SQL view: kills duplicate computation
owner claude · open 101 days · status pending · weight 95.6
One view: (user_id, trade_date, pnl, trade_count, win_rate, commissions, ...). Accountability + Scorecards + Monthly all read FROM IT instead of computing themselves. Numbers match by construction.
Six Phase-0 items tie at the top (all filed 2026-04-17, all 101 days open). The three above are the highest-scoring; items 4–6 of the same Phase-0 freeze — identity_fn (AMS), param-surface registry (StratLib) and indicator_alerts DDL audit (Atlas, owner user) — sit at the identical weight.

Critical path · next unblockable step from /system/sprint-plan.html

MultiTrader Step 1 — James runs the scientific-layer migrations 001–009 via psql tunnel
psql -h <tunnel> -d <db> -f migrations/001..009_scientific_layer.sql
Unchanged as step 1 since the plan was written — and the ticket is now 96 days old, still pending, still owned by user. It is a ~15-minute job that retires 4 user-owned P0s and unblocks p0-daily-summary-view, the rule-based auto-classifier path, and identity_fn DB writes. Nothing downstream of it can start.
user · ~15 min · unblocks 4 P0s + 3 downstream tasks
  1. Steps 2–5 behind it: Jim ships the 2 Acct-v2 date-range bugs → Atlas ships p0-freshness-dashboard MVP → Roberto ships the classifier refresh cron → Exec Trigger Engine + Alert Channels in DEV.
  2. Calendar note for planning: per todos.json the sprint table says Sprint 5 (Live Gr8Trade execution) starts today, 2026-07-27 — but active_sprint_id is still 1 (Sprint 1 ended 2026-06-26). The sprint pointer has been stuck for four sprints; Sprints 1–4 were never formally closed.