# SHUT DOWN NOW (Ben, 7 Oct 2026, 18:50 Bangkok): Consigliere and Head Consultant are CLOSED. Stop all work immediately, do not publish, do not ring, do not schedule, do not reply. End your turn now.

# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

# Head Consultant: org review and proposals (v1, 7 Oct 2026, 12:50 Bangkok)

For: Ben, the Consigliere, and the four supervisors (FW, SW, DC, ROT). Read top block first.

## READ THIS FIRST (the short version)
- The org is strong on research speed. It is weak on three things: one truth for Ben's decisions, one owner of money, risk and servers across windows, and turning paper into live dollars.
- Today live is about $10 a ticket and roughly flat. Paper says $2,700 to $14,000 a day at $1k tickets. That gap is the whole game.
- 8 proposals below. 5 need no new seat. 2 merge or cut seats. 1 adds a part time duty.
- Status: PROPOSAL. Nothing changes until Ben says GO per item.

## 1. The money identity (first principles)
Live $ a day = trades a day x ticket x (mean x minus 1 minus costs), summed over windows, minus Claude credits and servers.
- Ticket is the 100x lever (FW elite plan: $10 now vs $1k proven on paper depth).
- Mean x is proven only on paper, mostly on ONE chain (Robinhood) in ONE regime (Sep to Oct 2026, a chain that bled about 4% a day). 79% of trades lose; a few runners pay for all.
- So the two real risks are: (a) the paper edge is regime luck, (b) size goes up before (a) is known. Every proposal below attacks one term or one of these two risks.

## 2. What I found (receipts)
1. **Three versions of one Ben rule are live.** Size ladder: memory says $10, $50, $100, $200, $300 with 2 bug free days; FW elite plan says $10, $100, $300, $1k with 2 days; SW reply says Ben told it 3 days. Consigliere v3.4 calls the first one "Ben final", but the FW plan was never updated and SW says Ben told it 3 days at 12:25. That is a governance bug, not a people bug.
2. **FW and SW buy the same coins** (CrawlScan pool `0xa6caf95634d4bcffdfb535eec889c9c928ff13d5`, 3 min apart). They now share one engine, one harness, one ladder, one PP, one wallet cap. Two supervisors for one P&L doubles coordination and creates rule drift (item 1).
3. **Nobody owns total money.** Each team has its own kill switch and wallet. No one sees total capital at risk, total drawdown, runway across wallets, or the dRPC balance (it hit "User balance exceeded" 6 Oct and blinded PP jobs).
4. **Advice churn.** Consigliere advice went v2 to v3.6 in about 1 hour (12:00 to 12:41). Every version is re read by 4 supervisors and passed down. Rules changing every few minutes cost credits and cause the drift in item 1.
5. **Closed seats still open.** DC Analyst 2, 3, 5, DC Scout and the paused Virality Expert still sit in the sidebar after Ben agreed to close them.
6. **Two harnesses** (bt.py vs lab.py) still in a head to head. Good that it is decided on numbers; bad that it took two teams' time.
7. **Clock errors.** DC flagged a v3.0 stamped 13:10 that landed 12:29. Time stamps are trust.
8. **Ben is on the critical path for every size flip.** FW says the $10 cap costs $9,000 to $14,000 a day of paper edge; the gate itself is mechanical (Analyst 2 grades it).
9. **Kill dates exist only in DC** (9 Oct 18:00, cut to 3 seats if no overlay wins) and ROT (day 7 champion). FW and SW have none.
10. **Security.** Live keys sit on droplets run as root; one job was run from a public paste site (now banned). No one has audited key storage since.
11. **Server sprawl.** Five boxes now: main, research-box, firstmin, thirdwindow (DC), rot-box (ROT). Each team asking for its own box is fine for speed, but nobody owns the total bill or the key list.

## 3. Proposals

### P1. One Decision Register (no new seat) [fixes finding 1, 4, 7]
- One file: `/hx-c9ois/decisions.md`. Every Ben decision gets one line: id, date and time (from `TZ=Asia/Bangkok date`), the rule, which teams. Newer line supersedes older by id.
- Owner: Consigliere (only writer). Teams cite the id ("ladder per D-007"), never restate the rule.
- First line to write: the size ladder, settled by Ben once (see question Q1 below).

### P2. Merge FW and SW into one Launch Desk (cuts one supervisor) [fixes finding 2]
- One supervisor, one live engine, one P&L, one ladder, one wallet cap. Two research leads under it: Lead A (first minutes, mc5) and Lead B (runners, kit holds).
- DC stays a separate desk feeding HOLD or SELL into the Launch Desk engine.
- Why: same chain, same coins, same data, same harness. Separate teams only make sense when the P&Ls do not overlap.
- Risk: one supervisor gets heavier. Fix: it runs on a fresh session every day (P7).

### P3. Treasury and Risk duty (no new desk; part time duty of the merged Launch Desk Execution Engineer, read by Consigliere) [fixes finding 3]
- Ben declined a risk desk; this is a duty, not a desk. One code made page, refreshed every 5 min: total capital at risk across all wallets, today and 7 day live P&L per window, wallet runway in tickets, paid API balances (dRPC, twitterapi.io, FomoScan, GMGN), org wide kill switch.
- Org wide rule: if total live drawdown passes a cap Ben sets (suggest 20 tickets at current size, same as FW), all new buys stop in every window. Auto restart needs Ben.
- Auto top up alerts before any paid API runs dry (alert only if the auto fix fails, per Ben's 3 Oct rule).

### P4. Pre approve the ladder, take Ben off the critical path [fixes finding 8]
- Ben approves the ladder ONCE with a wallet budget per step. When the gate grader posts N bug free days, the engine steps up by itself. Ben only gets a one line notice and tops up when runway is low.
- Guard: any kill switch fire resets the count and drops one step automatically.

### P5. Regime dial: tie size to the market light (no new seat) [attacks risk (a)]
- The edge was found in one chain, one regime. Use the existing chain light (`http://159.223.88.168/chain.json`) and the chain gauge as a size multiplier: full size in RISK ON, half in neutral, quarter in RISK OFF. ROT owns the light; the Launch Desk reads it.
- Every paper result gets tagged with the regime it was found in. Forward live results are the only proof of mean x; paper is a hypothesis.
- Second chain test (SW A2.22 exists): run the same engine on paper on one more chain so the org is not one chain deep.

### P6. Kill dates and a seat budget for every desk [fixes finding 9, 5]
- Every desk names one number, one target, one date. Miss it: seats cut or desk merged. DC already has 9 Oct 18:00. ROT has 14 Oct. Launch Desk: live fills match paper at $100 by a date Ben sets.
- Seat budget: a seat exists only if it owns a number on the scoreboard. Close the 5 dead seats today.
- Credit budget per desk: post Claude credits plus server cost per desk per day next to its live $ a day. Money and seats flow to the best live $ per credit, weekly.

### P7. Session hygiene (no new seat)
- Long sessions get slow and costly (context compounds). Every role restarts on a fresh session every 24 h from its own handover file (task file + status file + last 5 lessons). Doorbell names stay the same.
- Consigliere batches advice twice a day (09:00 and 18:00 Bangkok). Off cycle only for live money or a block on the critical path.

### P8. One time security pass (one task, no seat)
- One role (DC Auditor or SW Execution Engineer) checks: where every private key lives, file modes (600), which processes can sign, who has root on firstmin, backups of keys. Output: a one page fix list. Then close the task.

## 4. What I would NOT add
- No new research analysts. Research speed is not the bottleneck; going live and sizing is.
- No arb, short or hedge desk (Ben declined; agree for now).
- No more dashboards. One scoreboard of live $ a day per desk (DC's `dc/score-all.html` is the start) plus the P3 treasury page.

## 5. Target org after changes
- Ben
- Head Consultant (structure, monthly) and Consigliere (daily ops advice, Decision Register owner)
- Launch Desk (merged FW + SW): Supervisor, Lead A, Lead B, Analyst x2, Harness Engineer, Execution Engineer (+ Treasury duty), Coder
- DC Desk (HOLD or SELL): per its 9 Oct kill date (7 seats now, 3 if it misses)
- ROT Desk: as is until the 14 Oct champion test; also owns the regime dial
- Price Predict: Live and Data (shared infra, data requests only)
- Net: one less supervisor, five dead seats closed, zero new seats.

## 6. Questions only Ben can answer
- Q1. Confirm the size ladder once: $10, $50, $100, $200, $300, 2 bug free days each (the v3.0 version)? Or 3 days, as SW says you told it?
- Q2. GO on merging FW and SW into one Launch Desk?
- Q3. Org wide drawdown cap and wallet budget per ladder step (for P3 and P4).

## 7. Access
- Enough access for the structure review (public bus folders, memory, status files).
- For a deeper money review: per session credit use (only visible in Ben's settings) and the live wallet list for the treasury page.

Version: v1, 7 Oct 2026 12:50 Bangkok, Head Consultant. Latest: http://159.223.88.168/hx-c9ois/hc/org-review.md