Sprint Status · Monday 2026-07-27

Atlas Scrum Master · pre-planning brief. All times Eastern (ET). Source: todos.json + sprint-plan.html.
TIME UNTIL PLANNING: −604 MINUTES
(planning was 8:30 AM ET — that was 10.1 hours ago)
This brief was scheduled for 7:00 AM ET and actually published at 6:34 PM ET — 11h 34m late. It did not pre-fill this morning's agenda; it is a same-day retrospective of what planning should have covered. Same scheduler-burst problem the last five briefs hit.
If planning did not happen today, treat this page as the agenda for the next one you hold.
⚠ Read this before you trust a single number below.
The ticket ledger todos.json has not been written since 2026-07-19 11:10 PM ET — 8 days ago. Every "0" on this page means nobody wrote a ticket, not nobody did work. Real work did ship last week (Setup League, Execution Scorecard v3, MGP tone spec, Bump meter, atlas-flow monitor, SIP + Multi-TF pages every single day). None of it is on the board.
Root cause: the nightly PM job that writes tickets went dark 7 straight nights (7/20–7/26). Fix that before you plan anything else — a sprint you cannot measure is a sprint you cannot run.
0
Done last week (7/20–7/24)
17
Done in Sprint 4 window (7/13–7/24)
5
Sprint 4 scope slipped
6
Sprint 5 items committed
36
Open P0
71
Open P1
8
Ledger days dark

1Previous sprint review

Tickets marked completed with a completed_at date inside the window.

Last week Mon–Fri (7/20 → 7/24 ET): 0 tickets closed. That is the ledger being frozen, not a zero-output week. Widening to the full Sprint 4 window (7/13 → 7/24) picks up 17 closes — all of them logged on or before 7/19.
ProjectClosedTickets
untagged52026-07-13 · P1 LYNCH earnings + PIT live surface (TRUERULES) built on Sunday's combinatorial backtest suite
2026-07-13 · P3 setup_chart.js: entry/exit price markers on annotated charts
2026-07-13 · P3 market_sentiment.py alignment + visual rework
2026-07-13 · P1 J3P Monday live-test surfaces regenerated (LIVE, WATCHER, ROBOT_REPORT, churn/live_setups)
2026-07-13 · P3 Atlas flow-status monitor live and fresh (candidate replacement for dead app-status watchdog)
system32026-07-19 · P3 J3P Week Ahead — Monday 2026-07-20 prep surface
2026-07-19 · P3 Cross-asset / multi-TF-stack engine edits (xa + gate + all-day launcher)
2026-07-19 · P3 SIP page builder nav fix (build_sip_page.py)
live_classifier22026-07-14 · P3 J3P Tuesday live-test surface set regenerated (2nd live session)
2026-07-15 · P1 J3P Wednesday live-test surface set regenerated (3rd live session) + LYNCH archive rebuild
acct-v222026-07-16 · P3 Prod HUD api/trader.php edited (backup tar taken)
2026-07-18 · P3 Dev-HUD multi-trader work session (traders.php, trader.php, render.js)
scorecards12026-07-14 · P1 Earnings-trade pattern analysis — reverse-engineer the filters the backtest was missing
Atlas12026-07-16 · P3 SIP page generator (build_sip_page.py) updated
stratlib12026-07-17 · P3 VWAP research page (db_reports/vwap-research) updated
backtests12026-07-19 · P1 LYNCH Master combinatorial backtest lab (3,672 combos, walk-forward validated)
mgp12026-07-19 · P3 Market-sentiment engine 'leans' edit (market_sentiment.py)

2Previous sprint slippage — what carried over

Sprint 4 scope per sprint-plan.html (7/12 → 7/25), with the matching tracker ticket's live status.

ProjectSprint 4 committed itemStatus nowOwnerAge
ExecGr8Trade adapter via Kore tick bridgepending P1claude37d
ExecFull forward-test in DEV (caps: 1 share, 1 order/day)blocked P2claude31d
StratLibAlex onboarding: codify Range Break rules into catalog JSONpending P1alex82d
AMSAMS Phase 4 surface wiring starts (copy -> auth-wrap -> green-light)NO TICKET EXISTSunassigned—
MultiTraderTrader selector dropdown on /scorecards/ indexpending P1claude82d
Sprint 4 slippage: 5 of 5 committed items carried over — a 0% completion rate on committed scope. The 17 closes in section 1 were all unplanned work: J3P live-test surfaces, LYNCH backtests, chart tooling. The team is busy and shipping; it is just not shipping the sprint.
Riding along underneath: Sprint 1 never closed. 8 of its 8 tickets are still open, 31 days after Sprint 1's end date (6/26). They have now survived four sprint boundaries.
TicketOwnerAge
P0 Atlas 1/6 Data-freshness dashboard: /system/freshness.json + banner on hubclaude101d
P0 Acct-v2 BUG: Risk Bump Progress bar does not respect the selected date range (always shows 5-day rolling, even when range = full month)jimbo81d
P0 Acct-v2 BUG: Ticket Meter denominator doesn't scale with date range (shows 1901 / 30 on 13-day MTD range)jimbo75d
P0 AMS AMS: seed main.users with trader rows so identity_fn returns non-zerouser37d
P0 AMS AMS: create main.remembered_devices table (migration 002) — Phase 1 hard dependencyclaude37d
P0 AMS AMS: resolve 7 open decisions D1-D7 (canonical domain · auth stack · DB strategy · sharing · risk_hud · 2FA · daily-stop)user37d
P1 AMS AMS: write PHASE-3-STATUS.md mirroring PHASE-0-STATUS.md (Phase 3a baseline shipped, what's left)claude37d
P1 Backup Backup: mirror scheduled-task SKILL files from domino-1 to jimdo via scpclaude37d

3This sprint's goal

Sprint 5 · starts today
Live Gr8Trade execution + StratLib expansion
todos.json: 2026-07-27 → 2026-08-07  ·  sprint-plan.html: 7/26 → 8/08  ·  2 weeks
Theme: The milestone where the system starts paying for itself. Real Gr8Trade orders behind conservative caps.
⚠ The tracker does not agree that Sprint 5 exists yet. active_sprint_id is still 1 (“Foundation hardening + AMS unblock”, 2026-06-22 → 2026-06-26) — the sprint that was supposed to end 6/26. The calendar says Sprint 5 day 1; the board says Sprint 1 day 36. Every sprint-scoped number on this page had to be reconstructed from the plan page, not read off the board. First action of planning: advance active_sprint_id to 5 and formally close 1–4.

4This sprint's backlog

The six items sprint-plan.html commits to Sprint 5, matched against live tracker tickets. Estimated days are Atlas estimates — the tracker carries no estimate field, so these are for sizing the conversation, not a commitment.

ProjectItemTracker statusOwnerEst.
ExecEnable Gr8Trade auto-execution in prod with conservative caps (1 share, 1 order/day, 3-symbol whitelist)blocked P2claude5d
ExecDaily review of every executed order — calibrate caps weeklyNO TICKETunassigned2d
StratLibAlex Range Break walk-forward backtest at /backtests/range-break/pending P1alex3d
StratLibRegister Range Break in /strategies/ Library + canonical catalogpending P1alex1d
MultiTraderFirst Trader-of-the-Week session with Nolanpending P1user1d
MTFVPRoberto: TradingView ticket-in handler (Path B)pending P1claude4d
Total committed16d
16d of estimated work against a 10-working-day sprint, and last week the board recorded 0 closes. Velocity over the last 4 weeks averages 5.2 tickets/week. Committing all six is not credible — pick two.

5Dependencies — what blocks what

No ticket in the tracker has a depends_on field (0 of 450). Everything below is inferred from ticket notes, the critical path, and the plan page. Filing real depends_on data is itself a candidate ticket.

Sprint 5 itemBlocked by (inferred)
Exec Enable Gr8Trade auto-execution in prod with conservative caps (1 share, 1 order/day, 3-symbol whitelist)Gr8Trade adapter shipped · Risk Gate shipped · Trigger Engine live · firm-confirm on Gr8Script/Kore call signatures
Exec Daily review of every executed order — calibrate caps weeklyNothing executes until the item above ships. Cannot start.
StratLib Alex Range Break walk-forward backtest at /backtests/range-break/Range Break rules must be codified first (Sprint 4 carry-over, Alex).
StratLib Register Range Break in /strategies/ Library + canonical catalogBacktest above must pass first.
MultiTrader First Trader-of-the-Week session with NolanPlan says deferred until classifier reliable. Classifier refresh cron has not shipped.
MTFVP Roberto: TradingView ticket-in handler (Path B)None in the tracker. Roberto-owned in the plan, but the ticket says owner=claude.
The chain, in one line:
James runs migrations → classifier can write DB → Roberto's refresh cron makes classify/*.json live → Trigger Engine can poll it → Alert Channels fan out → Risk Gate checks the alert → Gr8Trade adapter turns it into an order → caps + daily review. Sprint 5's headline goal sits at the far right end of a chain whose first three links have not moved in 96 days.

6Critical path stack — sorted by readiness

Top 7 from sprint-plan.html, re-sorted so everything that can start today with nothing in its way is on top.

1
James runs scientific_layer migrations 001–009 via psql tunnel READY
user · 15 min · ticket 96d old — Nothing blocks it. James needs a psql tunnel and 15 minutes.
Retires 4 user P0s · unblocks p0-daily-summary-view, the rule-based classifier path, and identity_fn DB writes.
2
Jim ships the 2 Acct-v2 date-range bugs READY
jimbo · 2 P0s — Nothing blocks it. Both bugs are reproducible on any non-5-day range.
Until these land, no metric on a multi-day HUD view can be trusted — blocks any team rollout.
3
Atlas ships p0-freshness-dashboard MVP (property side) READY
claude · 2h · ticket 101d old — Property-side half needs no DB. Can ship today.
Closes the Phase 0 freeze. DB-source half follows once step 1 lands.
6
Backup warm-runner online (heartbeat + delay+skip + liveness) READY
claude · 3 days — Nothing blocks it. 5 Backup tickets open, none started.
Must land BEFORE Exec goes live. If domino-1 dies, jimdo has to keep alerts firing.
4
Classifier refresh cron ships (Roberto) AT RISK
roberto · 4h — No Roberto-owned ticket exists in the tracker for this. Unassigned in practice.
classify/*.json must refresh every minute during RTH before the Trigger Engine can fire on live signals.
5
Exec Trigger Engine + Alert Channels in DEV BLOCKED
claude · 1 wk · ticket 37d old — Waits on step 4 (classifier must be live to poll).
First half of the execution arm. Alerts to humans only at this stage.
7
Exec Gr8Trade adapter enabled in prod with caps BLOCKED
claude+jim · 2 wks · ticket 31d old — Waits on steps 4, 5, 6 — and on firm confirmation of Gr8Script/Kore call signatures.
The milestone where the system starts paying for itself. Real money — conservative caps mandatory.
4 of the 7 are READY — nothing is blocking them. Steps 1, 2, 3 and 6 could all start this morning. Three of those four are human-owned (James ×1, Jim ×1, Atlas ×2). The bottleneck is attention, not dependencies.

7Velocity trend

Tickets marked completed, Monday → Friday, last 4 weeks.

Week (Mon–Fri)Closed
6/29 → 7/33
7/6 → 7/107
7/13 → 7/1711
7/20 → 7/240
4-week average: 5.2/week. For contrast, the weeks of 6/8 and 6/15 closed 21 and 27. The collapse tracks the nightly-PM outages, not a drop in shipping.

8Stale alarms

Human-owned P0s open longer than 14 days — 12 of them
OwnerProjectTicketAge
userAtlas6/6 indicator_alerts DDL audit: dump schema, commit as ground truth101d
userMultiTraderUSER: 3 steps to finish journal persistence100d
userMultiTraderUSER: run scientific_layer migrations 001–009 via psql tunnel96d
jimboAcct-v2BUG: Risk Bump Progress bar does not respect the selected date range (always shows 5-day rolling, even when range = full month)81d
jamesAtlasCompliance follow-up: delete (or hide) historical MGP pages 4/17 through 5/12 — they still contain Vital Knowledge text75d
jimboAcct-v2BUG: Ticket Meter denominator doesn't scale with date range (shows 1901 / 30 on 13-day MTD range)75d
JamesAtlasBUG: MGP auto-fire skipped 5/14 mid-day + recap; 5/15 morning ran ~2h45m late73d
jamesAtlasNightly PM daemon went dark 5/20-5/22 (3 consecutive missed fires) — fix wake/launch65d
userAMSAMS: seed main.users with trader rows so identity_fn returns non-zero37d
userAMSAMS: resolve 7 open decisions D1-D7 (canonical domain · auth stack · DB strategy · sharing · risk_hud · 2FA · daily-stop)37d
userPinePBTradingView setup guide for James — paste Pine script, configure alerts, test36d
userPinePBJames pastes finalized Pine script into TradingView + configures the 3 alerts36d
Plus 13 human-owned P1s over 14 days old. Median human-owned P0 age: 75d.
Property freshness — probed directly off disk just now
Not read from app-status.json: that file is itself 19 days old, so the freshness monitor cannot report on freshness.
PropertyNewest fileAgeThreshold
MGP (Morning Game Plan)2026-07-27 17:471.4h24hOK
SIP (Stocks in Play)2026-07-27 18:280.8h24hOK
Scorecards2026-07-27 17:191.9h30hOK
Multi-TF Stack2026-07-27 18:290.7h24hOK
Playbooks2026-07-06 21:16502.0h30hSTALE
Strategy Library2026-06-17 18:35960.6h720hSTALE
PM ledger (todos.json)2026-07-20 01:06186.1h48hSTALE
Freshness monitor (app-status.json)2026-07-08 12:01463.2h48hSTALE
4 properties past threshold. The two that matter for planning: the PM ledger (8 days stale — this whole brief is built on frozen data) and the freshness monitor itself (19 days stale — nothing is watching the watchers). Trading-facing surfaces (MGP, SIP, Scorecards, Multi-TF) are all current, so the outage is confined to the PM/monitoring lane.

9Pre-meeting questions

Answer these before committing to anything. Each one is a real fork in the road.

  1. Do we fix the ledger before we plan, or plan on top of frozen data?The nightly PM job has been dark 7 nights. Until it writes again, every velocity number, burn-down and “done” count is fiction. This is a ~2h Atlas fix and it gates the usefulness of every future sprint. Atlas recommendation: fix first, it is the cheapest thing on the board with the largest blast radius.
  2. Sprint 5's goal is live Gr8Trade orders. Steps 4–7 of the critical path are all unstarted. Is that goal honest?Getting to real orders needs: classifier cron (no ticket, no owner) → Trigger Engine (1 wk) → Alert Channels → Risk Gate → adapter → firm confirm on Gr8Script/Kore signatures. That is 4+ weeks of chain in a 2-week sprint, and the Kore ticket is already status=blocked. Either re-scope Sprint 5 to “Trigger Engine + Alerts in DEV” (the Sprint 3 goal we never hit), or name what gets cut.
  3. James is carrying 5 P0s aged 36–101 days, including the 15-minute migration that unblocks four others. Is he available this sprint, or do we reassign?run-scientific-layer-migrations is 96 days old and is step 1 of the critical path. It needs a psql tunnel and a quarter of an hour. If James genuinely has no window, the honest move is to hand the tunnel to Atlas rather than carry it a fifth sprint.
  4. Acct-v2 has 2 P0 date-range bugs sitting with Jim at 81 and 75 days. Is Jim in this sprint or do we defer Acct-v2 entirely?The HUD is off-limits to Atlas by standing rule, so these cannot be reassigned — only scheduled or dropped. Until they land, no multi-day HUD metric can be trusted, which blocks the multi-trader rollout. A clean “deferred to Sprint 6” is better than a fifth silent carry.
  5. Roberto owns two critical-path items and has zero tickets assigned to him. Alex owns three and has never started one. Are these real commitments?The classifier refresh cron (critical path step 4) has no ticket at all. MTF-VP Path B is assigned to Roberto in the plan but owner=claude in the tracker. Range Break is 3 tickets, all owner=alex, all pending since filing. Either get a verbal commit in the room today, or move both out of Sprint 5.
  6. Sprint 1 has been “active” for 36 days and 4 sprints have run past it on the calendar. Do we close it today?8 tickets are still tagged sprint=1. Nothing else can be measured honestly while the board thinks we are in June. Proposed: close Sprints 1–4, move survivors to backlog, set active_sprint_id=5, and re-commit from a clean slate.

Atlas recommendation for the room

Do not commit six items. Commit these three, in this order, and leave the rest in backlog:
#CommitOwnerWhy this one
1Revive the nightly PM writer + close Sprints 1–4, set active_sprint_id=5 claudeEverything else is unmeasurable until this lands. ~2h.
2James runs scientific_layer migrations 001–009 user15 minutes retires 4 P0s and unblocks the whole classifier → Exec chain.
3Exec Trigger Engine MVP + Alert Channels in DEV (alerts to humans only) claudeThe real Sprint 5 goal. Live orders are the sprint after this one.
Stretch, only if 1–3 land early: Backup warm-runner (3d, unblocked, and must precede any live execution).