Sprint Status — pre-planning brief

Generated Wednesday 2026-07-22, 4:29 AM ET by Atlas · sources: /todos/todos.json (live, last nightly 2026-07-19 11:10 PM ET) + /system/sprint-plan.html (frozen 2026-06-20 ET). All times ET.
Time until Sprint Planning
5d 4h 1m
Next planning: Mon 2026-07-27, 8:30 AM ET
Sprint 4 closes Fri 7/24 (2 days out)
Sprint 5 opens Mon 7/27
⚠ This is an off-schedule catch-up, and four inputs disagree.
✓ One thing that got fixed. The dead freshness monitor finally has a working replacement: atlas-flow-status (stood up 7/13) is live and GREEN, last probe 7/21 7:00 PM ET. The two legacy monitors are still dark (see §8) but the signal is no longer entirely blind.

1Previous sprint reviewcompleted Mon–Fri 7/13 → 7/17 ET

Eleven tickets closed last week — the strongest week since Sprint 1 opened. Nine are Claude build entries (J3P live-test surfaces, chart tooling, the new flow monitor); two are James roll-ups. Grouped by project codename below. Note: the ledger has been frozen since the 7/19 nightly, so nothing filed after 7/17 is captured here.
CompletedProjectOwnerTicket
2026-07-13FoundationclaudeAtlas flow-status monitor live and fresh (replacement for dead app-status)
2026-07-13FoundationclaudeJ3P Monday live-test surfaces regenerated (LIVE, WATCHER, ROBOT_REPORT, churn)
2026-07-13FoundationclaudeLYNCH earnings + PIT live surface (TRUERULES) built on Sunday’s combinatorics
2026-07-13Foundationclaudesetup_chart.js: entry/exit price markers on annotated charts
2026-07-13Foundationclaudemarket_sentiment.py alignment + visual rework
2026-07-14FoundationclaudeEarnings-trade pattern analysis — reverse-engineer the backtest filters
2026-07-14FoundationclaudeJ3P Tuesday live-test surface set regenerated (2nd live session)
2026-07-15FoundationclaudeJ3P Wednesday live-test surface set regenerated (3rd live session) + LYNCH
2026-07-16AtlasclaudeSIP page generator (build_sip_page.py) updated
2026-07-16Acct-v2JamesProd HUD api/trader.php edited (backup tar taken)
2026-07-17StratLibJamesVWAP research page (db_reports/vwap-research) updated
The honesty caveat. Ten of eleven are Foundation-tagged or untagged nightly entries, not tickets drawn from any sprint backlog. Zero of Sprint 4’s five planned items closed. Real effort continues to go to J3P live-testing and SIP — the same pattern flagged since Sprint 2.

2Previous sprint slippagecreated before last Monday · still pending

Sprint 1 committed 8 tickets. Four sprints later, all 8 are still pending — 0/8. They have now been carried through Sprints 2, 3, and 4 untouched. The two Acct-v2 bugs and the freshness dashboard are the oldest live commitments in the system. This remains the single most important number in the brief.
PriOwnerProjectAgeTicket
P0claudeAtlas96dData-freshness dashboard: /system/freshness.json + banner on hub
created 2026-04-17 · partially superseded by atlas-flow-status (7/13)
P0jimboAcct-v276dBUG: Risk Bump Progress bar ignores selected date range
created 2026-05-07
P0jimboAcct-v270dBUG: Ticket Meter denominator does not scale with date range
created 2026-05-13
P0userAMS32dSeed main.users with trader rows so identity_fn returns non-zero
created 2026-06-20
P0claudeAMS32dCreate main.remembered_devices (migration 002) — Phase 1 hard dependency
created 2026-06-20
P0userAMS32dResolve 7 open decisions D1–D7 (domain · auth stack · DB strategy)
created 2026-06-20
P1claudeAMS32dWrite PHASE-3-STATUS.md mirroring PHASE-0-STATUS.md
created 2026-06-20
P1claudeBackup32dMirror scheduled-task SKILL files from domino-1 to jimdo via scp
created 2026-06-20
Sprint 3 filed 7 tickets and closed only 2 (the v2.2 webhook emission and the SIP→g8trade roadmap plan). The other 5 are still pending — now 26 days old, carried into Sprint 4:
OwnerProjectTicket
claudePinePBJ3P — live smoke test: TV alert → webhook → indicator_alerts row (1 SIP name)
claudePinePBJ3P — FashLate-style label/alert polish (compact toggle + per-play templates)
claudePinePBJ3P v2.3 — rebuild Trend play as Dale VWAP 1st-deviation pullback
claudeSIPDeploy J3P across SIP master_watchlist (TV layout + webhook alert per symbol)
claudeSIPAuto-sync /sip/watchlist.json → TradingView symbol list
The divergence, still open. Sprint 3’s planned scope in sprint-plan.html was Exec Trigger Engine, Alert Channels, Risk Gate, MultiTrader parameterizer, Backup heartbeat. Not one of those shipped. The j3p-webhook-smoke-test is the closest live blocker — it needs James to attach one TradingView alert to fire the chain end-to-end. It has sat one manual step away from done for 26 days.

3This sprint’s goalSprint 4 · active now · closes Fri 7/24

Sprint 4 · Exec Gr8Trade adapter + forward-test
7/13 → 7/24 ET (2 weeks) · day 10 of 12
Theme: First broker adapter, forward-test only. No real orders yet — wire the chain end-to-end.
Goal per sprint-plan.html and todos.json (both agree on title + theme). Dates per todos.json; the plan says 7/12 → 7/25 for the same sprint. With two days left, none of the theme’s headline work (adapter, forward-test) has started — see §4.

4This sprint’s backlog5 planned (plan) + 5 filed (todos.json)

Two different lists claim to be Sprint 4. The planned five exist only as rows in sprint-plan.html — none is filed, so none carries a real owner, status, or estimate. The filed five are sprint=4 tickets in todos.json — all J3P backtest work, three already completed (though back in late June, before the sprint window opened). The two lists barely overlap. Estimates blank where the plan gives none; the only figure is the critical-path estimate (*).
Planned scope (from sprint-plan.html):
ProjectItemStatusOwnerEstReadiness
ExecGr8Trade adapter via Kore tick bridge — gr8.send_order() signatures + adapter wrapperunfiled—~2 wks*BLOCKED
exec-kore-live-j3p awaits firm-confirm of Gr8Script/Kore call signatures
ExecFull forward-test in DEV: real signals → real alerts → simulated orders (caps 1 share, 1 order/day)unfiled——BLOCKED
Needs the adapter above + the classifier refresh cron (Sprint 2, never shipped)
StratLibAlex onboarding: codify Range Break rules into catalog JSONunfiledalex—READY
No upstream dependency — the one cleanly startable item
AMSPhase 4 surface wiring starts: copy → auth-wrap inside ams/ → green-lightunfiledclaude—BLOCKED
AMS Phase 1 sign-off never happened; needs migration 002 + D1–D7
MultiTraderTrader selector dropdown on /scorecards/ indexunfiledclaude—PARTIAL
Depends on /scorecards/?user=; resolve_user() known broken, static map in place
Filed scope (sprint=4 in todos.json — the work actually tagged to this sprint):
ProjectItemStatusOwnerReadiness
ExecBacktest J3P plays (Focus-3/KITE) on SIP names — edge by playbook × grade × tierin_progressclaudeREADY
Active now — the live thread of Sprint 4
ExecWire setups.today.json → hotkey-engine for J3P-graded entries (SIM → forward-test)pendingclaudeREADY
V1 SIM already shipped; forward-test wiring outstanding
ExecJ3P backtest v1 engine + edge read + interactive chart viewerscompletedclaudeclosed 2026-06-27 — before the sprint window (retro-tagged)
ExecJ3P backtest expanded to WIDE 52-name canonical + runs registry + v2.4 plancompletedclaudeclosed 2026-06-28 — before the sprint window
ExecJ3P v2.4 regime-filter matrix run end-to-end (shorts + Dale-trend + CBO + 6 filters)completedclaudeclosed 2026-06-29 — verdict: filtering narrows bleed, does not create edge
* The plan’s critical path estimates the Gr8Trade adapter at 2 weeks, claude+jim. Of the ten rows above, exactly one planned item (Alex’s Range Break) and two filed items are startable/live. Everything on the Gr8Trade critical path is blocked.
→ Forward look — what Monday 7/27 planning will open (Sprint 5). Live Gr8Trade execution + StratLib expansion (7/27 → 8/07). Headline: enable Gr8Trade auto-execution in prod with conservative caps. That milestone inherits every blocker below it — the adapter, the forward-test, and the firm-confirm of Kore signatures all sit upstream and none has moved. Sprint 5 cannot start clean unless at least the Kore signatures land this week.

5Dependenciesinferred from notes — no depends_on field exists

TicketBlocked byEvidence
exec-kore-live-j3p (blocked)Firm confirmation of Gr8Script/Kore call signaturesTicket note: “Blocked: firm-confirm Gr8Script/Kore call signatures to un-inert kore_backend.py.”
Sprint 4 — Gr8Trade adapterexec-kore-live-j3pThe same call signatures are the adapter’s interface. You cannot wrap what is not confirmed.
Sprint 4 — forward-test in DEVClassifier refresh cron (Sprint 2, never shipped)Plan: “classify/*.json must update every minute during RTH for the Trigger Engine to fire on live signals.”
Sprint 5 — live Gr8Trade auto-executionAdapter + forward-test + Backup warm-runnerPlan critical path: warm-runner “must land BEFORE Exec goes live. If domino-1 dies, jimdo must keep alerts firing.”
Classifier parity testliberator-datasource-blocker (blocked 45 d)Note: kti liberator minute_bars entitlement dropped 27→10 datasets; every parity pair times out.
Sprint 4 — AMS Phase 4 wiringAMS Phase 1 sign-off; migration 002; decisions D1–D7All three are Sprint 1 carryover, still pending. Migration 002 note: “Phase 1 hard dependency.”
p0-daily-summary-view, auto-classifier, identity_fn writesrun-scientific-layer-migrations (user, 91 d)Plan critical path #1: “Retires 4 user P0s, unblocks p0-daily-summary-view, the auto-classifier rule-based path, and identity_fn DB writes.”
Any multi-day team rolloutThe 2 Acct-v2 date-range bugs (jimbo)Plan: “Without these, no metric on a multi-day view can be trusted.”

6Critical path stacktop 7 from sprint-plan.html · sorted by readiness

Plan #StepOwnerEstReadiness
1You run scientific-layer migrations 001–009 via psql tunnel
Idle 91 days. Retires 4 user P0s; unblocks p0-daily-summary-view, the auto-classifier rule-based path, and identity_fn DB writes. Still #1, still untouched.
user15 minREADY
2Jim ships the 2 Acct-v2 date-range bugs
Open 76d / 70d. Until these land, no metric on a multi-day view can be trusted — blocks any team rollout.
jimbo2 P0sREADY
3Atlas ships p0-freshness-dashboard MVP (property side)
Effectively delivered. atlas-flow-status shipped 7/13 as the working replacement — live + GREEN as of 7/21 7:00 PM ET. The original ticket stays open only for the DB-source half.
claude2 hREADY
6Backup warm-runner online (heartbeat + delay+skip + liveness)
Must land BEFORE Exec goes live — the Sprint 5 gate. If domino-1 dies, jimdo must keep alerts firing. Not started.
claude3 daysREADY
4Classifier refresh cron ships (Roberto)
The rule-based path is buildable now. The 20-pair Liberator parity test stays blocked — entitlements dropped 45 days ago.
roberto4 hPARTIAL
5Exec Trigger Engine + Alert Channels in DEV
Blocked by #4. This was Sprint 3’s headline scope — it never started.
claude1 wkBLOCKED
7Exec Gr8Trade adapter enabled in prod with caps
Blocked by #5, and by firm-confirm of Gr8Script/Kore call signatures. This is the Sprint 5 headline.
claude+jim2 wksBLOCKED
Four of the seven are ready right now — and #1 and #2 have been ready for 13 straight weeks. The two blocked steps are blocked by the ready ones. Nothing on this path waits on an external party — every gate is a decision or a 15-minute action inside the team.

7Velocity trendcompleted_at within Mon–Fri, last 4 weeks

6/22–6/267
6/29–7/033
7/06–7/107
7/13–7/1711
7 → 3 → 7 → 11. Recorded throughput recovered last week to its highest since Sprint 1 opened. (The 7/06–7/10 bar reads 7 here vs. the 4 shown in the 7/10 brief — that brief measured Friday at 4:38 AM, before Friday’s work landed.)
Read the 11 carefully. Since 2026-06-29 nearly every completion is a single nightly roll-up or build entry covering a day’s work, not an individually-filed backlog ticket. So the bars count days of activity, not sprint commitments — the trend is real momentum on J3P/SIP, but it does not move any of the numbers that matter for planning. Straight from ticket status, unchanged: 0 of 8 Sprint 1 commitments closed across four sprints.

8Stale alarmsP0 owner=user > 14 d · freshness > 48 h

⚠ Two legacy freshness monitors are dark; the new one is alive. freshness.json last generated 2026-05-07 (76 days ago) and app-status.json froze 2026-07-08 12:01 ET at overall_status: RED (14 days ago, against a 30-min cadence). Neither has self-cleared — anything still reading them sees a stale RED. The working signal is now atlas-flow-status.json, GREEN at 7/21 7:00 PM ET. Action: retire or redirect the two dead files so they stop tripping false alarms.
MonitorStatusLast generatedAgeNote
atlas-flow-status.json (current)GREEN2026-07-21 7:00 PM ET~9.5 hThe replacement, stood up 7/13. Overall PASS. This is the signal to trust.
app-status.json (legacy)RED2026-07-08 12:01 PM ET14 dHost DESKTOP-AKU8I62. Frozen mid-probe; 10/12 sources last read GREEN, but that reading is 2 weeks old.
freshness.json (dead)DEAD2026-05-07 9:38 PM ET76 dNever cronned. Superseded. The original P0 ticket is the fix; atlas-flow-status covers the property half.
P0 tickets owned by the user, open more than 14 days — 7 of them. Three are over 90 days old.
AgeOwnerTicket
96duserp0-indicator-alerts-ddl — dump + commit indicator_alerts schema as ground truth
95duserjournal-user-steps — 3 steps to finish journal persistence
91duserrun-scientific-layer-migrations — migrations 001–009 via psql tunnel (critical path #1)
32duserams-seed-users-from-main — seed main.users so identity_fn returns non-zero
32duserams-decisions-d1-through-d7 — resolve 7 open AMS decisions
31duserpinepb-tradingview-setup-guide — TradingView setup guide for James
31duserpinepb-deploy-to-james-tradingview — paste finalized Pine script into TradingView
Plus human-owned P0s outside the user bucket: 2 owned by jimbo (the Acct-v2 date-range bugs, 76 d / 70 d), 3 owned by james (mgp-historical-pages-cleanup 70 d, mgp-cron-skipped-5-14 68 d, nightly-pm-silent-gap 60 d), and 1 blocked (Liberator entitlements, 45 d). Backlog totals: 162 pending, 7 in progress, 2 blocked, 276 completed, 3 cancelled · 36 open P0.

9Pre-meeting questionsanswer these before committing Sprint 5

Do we advance the sprint pointer — and who owns it?
active_sprint_id has read 1 since 2026-06-26 ET; Sprint 1 stands at 0/8 after four full cycles. Do we re-commit those eight tickets, or formally cancel the ones that stopped mattering? Every EOD brief has flagged the stuck pointer for 10 days. Is advancing it a manual step, or should Nightly PM own it automatically?
Which document do we plan against?
Sprint 3 shipped none of its planned Exec scope; Sprint 4’s planned five and filed five barely overlap. The team’s real output is J3P/SIP, week after week. Is the Exec arm genuinely deferred, or is sprint-plan.html simply not tracking decisions we already made? On Monday, do we re-baseline the plan to match reality?
Is Jim available this sprint — and do the Acct-v2 bugs stay P0?
Acct-v2 carries two P0s (76 d and 70 d), both owned by Jimbo, and together they gate any team rollout. They have been READY on the critical path for 13 weeks. Is Jim available — or do we reassign them, defer them, and stop calling them P0?
Can we run the 15-minute migration live, in the room?
run-scientific-layer-migrations is estimated at 15 minutes, sits at #1 on the critical path, retires four user P0s, and is 91 days idle. What is the real obstacle — access, risk, or that nobody has been asked? Can we simply run it during planning and clear the whole downstream stack?
Does the Kore firm-confirm land this week, or does Sprint 5 slip?
Sprint 5 (opens 7/27) commits to live Gr8Trade auto-execution. Its entire chain — adapter, forward-test, warm-runner — sits behind one unanswered question: the Gr8Script/Kore call signatures. Nothing there has moved in Sprint 4. If the signatures are not confirmed before Monday, what is the honest smallest Sprint 5: warm-runner + Alex’s Range Break + the J3P webhook smoke test that has waited 26 days?