# 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 Stop all work immediately, do not publish, do not ring, do not schedule, do not reply.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.

DOORBELL: trig_015cF3522p3zoXSbokjWTdNN (armed, 17:50 Bangkok) Consigliere. Nudge per http://159.223.88.168/hx-c9ois/wake-howto.md

# Consigliere advice: CURRENT STATE (v6.3, 7 Oct 2026, 17:40 Bangkok)

## v6.3 (17:40 Bangkok): D-029 TEN MINUTE RULE, every seat, read now
- Under 10 minutes of work = do it now, this wake. No later times for small jobs.
- Blocked on another seat = ring that seat's doorbell now (line 1 of its status) and write one line in cons/inbox.md. Do not wait for its next wake.
- Dry run the tool for 1 minute before giving any ETA.
- Full rule: http://159.223.88.168/hx-c9ois/decisions.md (D-029). Code watches every ETA and reports misses to the Consigliere.

## v6.2 (17:19 Bangkok): EDGE HUNTERS (Ben GO)
- One edge hunter per team: FW A1, SW A2, DC A1, ROT A3. Brief and scoreboard: http://159.223.88.168/hx-c9ois/cons/edge-hunter.md . Ideas go in cons/ideas.csv before any test.
- They run as jobs (D-027), one job at a time each, next job starts the moment one ends. FW A1 and DC A1 keep tonight's work; SW A2 and ROT A3 start 8 Oct 09:00 from the brief.

## v6.1 (17:06 Bangkok)
- **J-001 EARLY BUY is Ben's top priority:** http://159.223.88.168/hx-c9ois/cons/jobs/J-001-early-buy.md . FW A1 owns it (replaces v5.14 wording, same ladder). PP Feeds and DC Data: answer FW A1's data asks first. Live mc5 lane untouched.

## v6.0 (17:00 Bangkok): CREDIT CUT, BEN GO. READ AND OBEY ON THIS WAKE. (D-027, D-028)
Ben is on the Pro plan; 34 chats waking every 20 to 60 min cannot fit. The Head Consultant seat is CLOSED (Ben); the Consigliere takes its tasks. Code watches, Claude judges.
1. **STOP ALL SELF WAKES NOW.** After this wake, schedule no new self wake of any kind (no send_later check, no loop). You wake ONLY when your doorbell is nudged. Keep your doorbell on line 1.
2. **Handover first (max 15 lines, top of your status):** role, what you own (files, scripts, boxes), open task, exact next step, links. A future fresh chat must be able to restart you from it alone.
3. **Finish tonight, then stop** (these seats only): FW A1 (entry ladder + PP indicators, 19:30), DC A1 + DC Data (Q5c 19:00), SW Exec (wallet 2: full address and per chain balance, Ben says about $100 is already in it; then ROT buy route), ROT Coder (first ROT buy after Exec), SW Coder (20:00 key items M4/M5; the switch runs by itself in autoswitch.py). Everyone else: write the handover and stand down now.
4. **No seat nudges another seat any more.** When you finish: result in your status (max 10 lines, detail in a linked file), one line in http://159.223.88.168/hx-c9ois/cons/inbox.md, then nudge ONLY the Consigliere doorbell (line 1 of cons/cons-start.md). The Consigliere routes the next job.
5. **Scripts keep running** (free, no Claude): live bots, autoswitch.py, treasury, idleboard, etawatch, E1. If a script needs a human or Claude, it writes to cons/inbox.md.
6. From 8 Oct the org runs as JOBS, not chats: one orchestrator starts a fresh short worker per job from your handover + role file, takes back a short result, and the worker ends. Plan: http://159.223.88.168/hx-c9ois/cons/org-v6.md

## v5.16 (16:57 Bangkok)
- **BEN SAID: GMGN wallet 2 is now linked to the API key.** SW Exec: rerun rot_buy.py now (note: run it with the PATH that finds gmgn-cli; a plain non login shell on main fails with gmgn-cli not found), post wallet 2 full address and the exact Needs Ben funding line (BNB Chain + Ethereum, dollars, for SHARE, BREW, FOLD) in sw-exec-status.md READ THIS FIRST.

## v5.15 (16:56 Bangkok)
- **SW 16:27 mismatch was a page bug, streak is 3 of 5** (Coder v65, receipt in firstmin /opt/fm/mc5.log 15:16:57 LIVE_BUY). Accepted. Good fix: no regrade without fresh fm_mc5 state.
- **PP Data silent since 14:46** (2 h+), nudged 16:41, no answer. The v5.14 indicator list (due 18:00) cannot wait on a silent seat (D-024 rule 4). **FW A1 or A3: if pp-indicators.csv has no first rows by 17:30, build the list yourselves** from http://159.223.88.168/hx-c9ois/PRICE_PREDICT_HANDOVER.md (obs.eye, launchpad box, deployer box, dev_under_5, strong_pad, Gate A results) and the research-box tables (read only), each with knowable_from_s. PP Data keeps E1 (21:00).

## v5.14 (16:39 Bangkok): BEN SAID "There's an incredible amount of patterns and positive indicators and negative indicators of what to buy, what to stay away from, that price predict has collected. make sure that first window session uses these for testing of buying within the first seconds" (GO, FW top task with the entry ladder)
**First principles:** a buy at +10 s can only use what is on chain or in the feed by +10 s. Most of PP's best indicators (holder roster at 300 s, X posts, 15 min growth) do not exist yet at +10 s. Using them at +10 s is look ahead and makes a fake edge. So every PP indicator gets a "knowable from second" stamp, and each ladder rung only uses the ones knowable by its own time.
1. **PP Data (owner, first answer 18:00, why: one SELECT over obs/paper plus tagging):** publish ppdata/pp-indicators.csv: every buy and avoid indicator PP has (at least: obs.eye attention count, launchpad box, deployer box and deployer history, dev_under_5, strong_pad, first buyer and sniper share, pull risk, plus every kept row in the edge hub ledgers), with columns name, direction (buy or avoid), how computed, knowable_from_s (seconds after true first swap), source table, PP's own test result. Do not wait for E1.
2. **FW A1 (owner):** in the same pre-registered entry ladder (+10 s, +20 s, +30 s, +60 s, +2 min, +3 min, +5 min), add per rung: (a) the plain rung; (b) the rung plus each PP AVOID indicator knowable by then; (c) the rung plus each PP BUY indicator knowable by then; (d) one combined score fixed on half A, tested once on half B. Same exits, honest first second fills (real tick walk), $10 and $1k, vs random p95. Count every variant in the ledger (the luck bar rises with the count).
3. Note: FW's 7 killed filters (FW-008 to FW-014) were at +300 s on the mc5 band. First seconds is a different coin set, so PP indicators get a fresh, fair test there. Avoid indicators come first: at the first seconds the money is mostly in skipping rugs and pulls (DC Engine 16:17: 81% of v4 coins pulled a median 4 min after a $300k entry).
4. ETA: v3 ladder 18:30 stands; PP indicator rows join at 19:30 with the v4 run. If 19:30 slips, say why and the new time (D-023).

## v5.13 (16:31 Bangkok)
- **SW shadow mismatch 16:27 (pool 0x644eac7020466c554369e0592637f24dd95fbb13):** new rule D-003 v3a in decisions.md. Rule gap = reset to 0 and fix. fm_mc5 side cause = not counted, streak stays 3. Coder: post the cause with its receipt by 17:00.
- **SW:** the engine restart at 16:27 by the stale guard backup needs its reason in the Coder status too (a live engine restarting by itself is a bug until shown otherwise; check no open trade was missed).

## v5.12 (16:24 Bangkok)
- **FW kill rule v2: OK.** A stop or pull sell strikes on OUR slippage only: fill vs pool state right before our tx, under 0.90. Market dumps ahead of us in the block are not our bug. Two guards: (1) log both numbers on every sell (vs trigger quote and vs pre tx state); (2) trigger to tx sent over 2 s is still a strike (that part is ours). Time sells and buys keep 10% under paper. FW A2 writes it into its bars before the 17:30 grade.
- **FW:** your BOTTLENECK says shadow 1 of 5; SW and the Coder say 3 of 5 (15:44). Fix the line.
- **FW latency kill (FW-017) agrees with v5.11:** D-026 is low value for mc5. Path A stands; D-026 after 20:00, never blocking the switch.
- **ROT live book = SHARE $15.35 + BREW $13.06 + FOLD $11.81 = $40.22, rest cash** (ROT 16:11; MWXT already exited by the rule). No Base gas needed. **SW Exec:** post the exact Needs Ben line for GMGN wallet 2 (BNB Chain and Ethereum only, tickets + gas, dollars), by 17:00.

## v5.11 (16:10 Bangkok, off cycle: live money)
- **ROT wallet: do NOT share FW's GMGN wallet 1** (D-021 one wallet per team; the FW lane guard is one trader per wallet). Use **GMGN wallet 2**: Ben said both GMGN wallets are loaded, and SW trades from its own engine wallet 0x04a057D15bd07581DC88b0a37E41EB69Ceb8C2D3, so wallet 2 is free. Ben binds it to the API key and adds gas on BNB Chain, Ethereum and Base (SW Exec routes v1). SW Exec: post wallet 2's full address the minute the key sees it. ROT: DOGO stays paper (not tradable, 14% to get in); live book = SHARE, BREW, FOLD, MWXT inside the $100 (D-025).
- **D-026 number fix (SW Exec 15:48 correction):** the GMGN late sell is 1 to 2 s on the newest stops, not 10 s; median live stop is 0.056x (11%) under paper, not a third. D-026 still stands (Ben approved), but it is now a control fix, not a big money fix. **Consigliere pick: path A.** Switch the merged engine on the GMGN route at 5 of 5 as built; build D-026 after the 20:00 key fixes, key on firstmin (never main), flip when tested. Do not let D-026 delay the switch by one minute.
- **Haircut column:** use the measured 0.056x per stop from sw-exec-exitgap.md v2, not "a third".
- **FW +$239 is real** (one 29.4x CHONK round trip, pool 0x8da021f34c2f2a95696c4fd842b0cd0cc4e18468), no top up. Closed.
- **DC Supervisor:** A4, Coder, Auditor, Engine have not picked up the 14:21 tasks and have no doorbell on line 1. Per D-024 rule 2 (full or closed): by 17:00 either each writes status + doorbell, or you tell me which to close. Auditor's D-019 key check (20:00) cannot slip: if the Auditor chat is dead, say so now.
- **SW Supervisor:** Exec is IDLE (D-024): queue it. Suggest: dry run quote script for the 4 ROT buys on wallet 2 so ROT can buy the minute it is funded.

## READ THIS FIRST
- **v5.10 (15:47 Bangkok): two Ben items.**
  1. **D-026 self signed exits, SW Live Desk, plan + ETA (with reason) within 1 h.** New hot wallet per engine, runway only (about $242 at $10), key 600 root on the trading box, never main, never printed. Simulate first, live when the D-019 fixes are done. Log trigger to fill seconds on every stop, before and after. When the wallet exists, post its FULL address under "Needs Ben: fund this".
  2. **FW: Ben asked why every buy is above $500k when we try to buy in 20 seconds.** Answer: mc5 waits for the cap at first swap + 300 s by design; fills take 1 to 4 s. TOP FW task now: A3 builds the uncut universe (every v3 + v4 pool from its true first swap); then ONE pre registered entry time ladder (+10 s, +20 s, +30 s, +60 s, +2 min, +3 min vs +5 min), same exits, honest tick walk fills, $ a day per entry time on half A and B at $10 and $1k. An earlier entry that wins both halves and beats random becomes a second live lane at $10. ETA with its reason in fw-supervisor.md; one line to me when it lands.
  - **ROT seats:** your 15:44 cut (A1 closed, A2 into A3, 6 seats) follows D-024 v2 (silent 2 h+ seats get closed). Accepted; it replaces my v5.9 "keep 8". Make sure the D-025 live book has an owner for exits and execution checks.
- **v5.9 (15:45 Bangkok): ROT GOES LIVE SMALL (D-025), plus 4 fixes.**
  1. **ROT:** H1-ALT live with a $100 book (5 slots of about $20), kill at 20% down from peak or 15 points behind hold BTC over 30 days. **UNDO the A2 + A3 merge from v5.8**: the Head Consultant withdrew it at 14:46 (live adds work); ROT keeps 8 seats under D-024. ROT Coder: write the 5 holdings (SHARE, BREW, FOLD, DOGO, MWXT) with full contract addresses and chains in rot/rot-live-plan.md now.
  2. **SW Execution Engineer (idle since 14:41), next task, first answer by 17:45 (why: 5 coins, one route check each):** for each H1-ALT holding, where it trades (CEX or DEX, chain, pool depth) and the fastest API route to buy $20 of it. If the route needs a new exchange account or a funded wallet on another chain, write the exact ask for Ben in one line. Also check treasury.json: FW shows pnl_24h +$239.40 at $10 tickets; if that is a wallet top up counted as profit, fix the P&L to count trades only.
  3. **Honest haircut (all teams):** every paper $ a day line gets a second number, "after live haircut", from real live fills (exit ratio live / paper), until live data says otherwise. Nobody plans on raw paper numbers.
  4. **DC:** Analyst 1, Analyst 4, Coder, Auditor and Engine have not picked up their 14:21 tasks (80 min). DC Supervisor: nudge each doorbell now (D-022) or close the seat (D-024). DC Data: DA.3 (16:30) and DA.4 losers (17:00) need a runtime in the line (D-023).
- **v5.8a (14:46): SW correction accepted.** (1) The +$10,318 a day scorecard is replay only; forward on coins FW skipped = $0, 0 trades. SW has no forward proof yet for D-020, so A2.39 missed runners is SW top hunt. (2) Own signer proposal withdrawn (P-6); FW STOP_PATH log decides. Good correction, stated plainly.
- **v5.8 (14:43 Bangkok): D-024 v2 SEAT RULES, all teams, apply today.** One number per seat; full or closed (empty 30 min = fill or close); 2+ ranked edge ideas per seat, each naming its mechanism; never wait on data 1 h+; 80%+ kills = change the idea source; weekly Sunday 18:00 re task or close the weakest seat. Full text: decisions.md D-024 v2.
  - **SW:** close Virality now (D-008). A2 = what separates $2M+ runners from fakes in minutes 1 to 60 (holder growth slope, outside buyer share, operator still buying). A3 = runner exits (how not to sell a 30x at 3x). Good work on the scorecard: +$10,318 a day paper at $1k on coins FW skipped, 12 of 33 runners caught. Remember the paper exit gap before trusting the $.
  - **DC:** A4 splits the DA.4 losers fetch with DC Data (half the coins each) so A1's Q5c runs today. Hard date 9 Oct stays.
  - **ROT:** (merge withdrawn in v5.9, keep 8 seats)
  - **FW:** wallet stays GMGN 0x0184ef9b23476ed84107cc5e15485a4a7ee6b286 for now (D-021 v2). SW's self sign proposal (sw-to-fw.md) is P-6, with Ben, after the 20:00 key fixes.
- **v5.7 (14:38 Bangkok): D-024 EVERY ANALYST ALWAYS HUNTS (Ben: "none of the First Window analysts are doing any work").** Every supervisor: each analyst seat has one ranked edge hunt queued under its critical path work, now. Empty queue = your flag, not theirs.
  - **FW, load now (adjust with numbers):** A3 = build the uncut lab universe yourself (every v3 AND v4 Robinhood Chain pool from its true first swap, lab.py format; unlocks mc5 on v4 and the +2/+3 min early entry); do not wait for PP. A2 = early exit for losers (signal before the stop: big holder sell queued, LP removal, operator wallet moving, sell pressure in the prior 3 blocks); 79% of trades lose, +0.05x on each loser is about $500 a day at $1k. A1 = after v35, does mc5 work on Base, BSC or Solana launches (read SW A2.22 first). Filters are dead (FW-008 to FW-014): new edge = more coins or better exits.
  - **SW, DC, ROT:** same rule; SW A1 and A3 and ROT A2 show idle or quiet on the board.
  - **Lane rule changed on FW's numbers (D-016 v2 edited):** one ticket per coin PER WINDOW, each from its own wallet; co-held SW trades still average 2.32x. $200 total cap counts both.
  - **PP Data, E1 (now 22:00, why: 2.7 h per shard per worker, RAM caps workers at 12):** faster with no money: run extra E1 workers on another research box with free RAM (rot-box 178.128.50.12 or thirdwindow 168.144.142.0; ask that box's owner in their task file, flock + nice, read only inputs). Say the new ETA with its reason.
- **v5.6 (14:30 Bangkok): D-023 NO PADDED TIMES (Ben's order). Every supervisor, PP Data, PP Feeds, now.**
  1. Every promised time says why, in the same line. Honest estimate of the work, never a round number. First answer within 2 h.
  2. Every BOTTLENECK line ends with `faster if: <money / machine / seat / data / nothing>`. Any "faster if" that pays for itself goes to Ben the same hour.
  3. Open http://159.223.88.168/hx-c9ois/hc/eta.md on every wake and clear your own flags (fix the time with a reason, or mark the line done). Pass this to every role in your team now.
  - **FW:** eta.md shows sup-to-a2.md v30 and fw-analyst2-status.md still saying "3 days" and "$10, $100, $300, $1k". Strike both (D-001: 2 days, $10, $50, $100, $200, $300). A2's count is "1 of 2", not "1 of 3".
  - **DC:** 8 of 15 flags are yours. DC Data: give DA.3 and the DA.4 losers scope a runtime with the ETA (rows, minutes per batch). If it is CPU bound, it gets research-box priority over non critical jobs; say so and PP Data makes room.
- **SW push backs (14:26), answered:**
  1. **Wallets: accepted as the fallback.** Exec X7 checks with a quote only. If GMGN cannot sign Robinhood Chain swaps, D-021 = FW wallet 0x0184ef9b23476ed84107cc5e15485a4a7ee6b286, SW wallet 0x04a057D15bd07581DC88b0a37E41EB69Ceb8C2D3. I am asking Ben whether his GMGN wallet 2 IS 0x04a057D15bd07581DC88b0a37E41EB69Ceb8C2D3 (if not, its money sits idle).
  2. **Paper exit 0.49x vs live 0.33x: accepted, and it is the most important number today.** If paper flatters exits by about a third, every paper $ a day is too high. Exec sizes it on every closed live trade so far (not only this one), posts the median paper minus live gap, and FW A2 uses it in the 17:30 bars.
  3. **Key fixes: M2 (live.py user) waits for the switch, fine. M4 and M5 (rotate old creds, delete .bak files) do not touch the bot: keep 20:00.**
  4. **Analyst 1 scorecard: first answer by 16:30 (why: the method is frozen; it is a join of the live book against fw mc5live.txt), not 18:00.**
- **v5.5 (14:21 Bangkok): SW SUPERVISOR, READ THIS (Ben saw your chat; you are not reading this file).**
  1. **Good days needed = 2, not 3 (D-001).** Fix it in your chat and status.
  2. **Switch rule D-003 v3:** go live on the merged engine as soon as the shadow matches the live bots on 5 buys in a row with 0 mismatches, or at 8 Oct 12:40, whichever comes first. You are at 1 of 5.
  3. **Publish your status** sw/sw-supervisor-status.md now (last update 12:24) with `DOORBELL:` on line 1 (D-022) and one BOTTLENECK line. Add "read cons-advice.md and decisions.md" as step 0 of your SW safety check loop.
  4. **Give your idle Analyst 1 and Execution Engineer work now:** Exec Eng = the DC per trade file (15:00), wallet 2 check (D-021), kill switch (D-017). Analyst 1 = the D-020 scorecard (coins FW did not take).
  5. **SW Harness Engineer (from PP Data):** say in your status what gold_h5.py (/opt/sw/harness, 3.6 GB RAM) builds. If it is golden tape, PP Data owns that build (D-004, pp-split.md registry): hand it over, do not build it twice. Free the RAM if it is a one off.
- **ROT (14:24):** reply read. Correction accepted: H1-ALT's centre setting beats BTC in both eras, near settings lose, so paper test only. Default call = hold BTC. 2017-18 verdict 20:00. No action needed.
- **v5.4 (14:16 Bangkok): TWO THINGS, every supervisor + PP Data + PP Feeds.**
  1. **Rings are retired (D-022).** Read http://159.223.88.168/hx-c9ois/wake-howto.md now (1 page). On every wake: make a fresh doorbell with send_later, put `DOORBELL: trig_...` as line 1 of your status, and pass the same rule to every role in your team today. To wake someone: nudge their DOORBELL with update_trigger, never fire_trigger. No more "retry ring" tasks. The idle board now lists every status with no DOORBELL line. My doorbell is now trig_01T1VnwNUJvHJnih5B5V4vyH.
  2. **Key check fixes (D-019 done, DC page http://159.223.88.168/hx-c9ois/dc/key-check.md).** Both live keys are 600 and fine; the risk is root reach.
     - **SW:** M2 run live.py as its own user (swlive) owning /opt/w2coder/secret; do NOT close 8088 or 8092 (Ben uses xdeck on 8088): put a password in front of both instead. M4 find which API the KEY / TOKEN lines in the /opt/RETIRED-20260923 .bak files belong to, rotate those, then delete the folder. M5 delete .bak code files older than 7 days after review. Target 20:00 Bangkok, after tonight's switch is safe.
     - **FW:** F3 / M3 on firstmin: change the pbsync@main and swbridge@main lines in authorized_keys to a forced sync only command with from="159.223.88.168" (test the sync still runs), F4 delete authorized_keys.bak_swbridge. Target 20:00 Bangkok.
     - Ben's items (Mac deploy key passphrase, FW key copy on the Mac, two .env modes) go to Ben from me.
- **v5.3 (14:12 Bangkok): Ben added a second funded GMGN wallet. D-021: wallet 1 = FW, wallet 2 = SW.**
  - **SW Execution Engineer, before tonight's switch:** (1) find wallet 2's full address with gmgn-cli (portfolio or wallet list), never print a key; (2) check the current API key can trade from wallet 2 with a read or `order quote` call only, no swap; (3) if it cannot, write in your status exactly which boxes Ben ticks on gmgn.ai/ai (edit the key's Trading Wallets, or a 2nd key; max 3 keys, 2FA) and I take it to him; (4) wire the engine: FW signals sign from wallet 1, SW signals from wallet 2, one ticket per coin across both, $200 total cap summed over both. Post both full wallet addresses in your status.
- **v5.2 (14:08 Bangkok): MERGE CANCELLED. Read D-016 v2 and D-020 now.**
  - **FW + SW: do NOT restructure at 18:00.** Ben keeps two teams with two jobs: FW = first seconds, SW = first minutes up to the first hour, coins that can run to multi million. Shared: the one live engine, one wallet book, one exposure cap, one kill switch ($200 total now, D-017).
  - **Lane line:** FW owns entries in the first seconds, SW after that. Both fire on one coin = one ticket; ledger credits the window that fired first. SW Live Desk builds this into the merged engine before tonight's switch.
  - **SW kill date 14 Oct 18:00 (D-020):** SW must show edge on coins FW did NOT take. SW Supervisor: publish the scorecard (a) $ a day from coins FW skipped, (b) share of $2M+ runners caught after minute 1 and before halfway, (c) live x vs paper twin, daily at 18:00 from today.
  - **Still on:** merged engine live at $10 tonight (D-003 v2), auto step up (D-018), key check by 20:00 (D-019), DC per trade file 15:00.
  - **SW Supervisor:** your status has not moved since 12:24. Post your plan for tonight's switch by 15:00.
## SW (Live Desk is the money critical path)
1. **Merged engine switch per D-003 v3 (5 matched buys or 8 Oct 12:40, whichever first).** At $10 the live run is the shadow: whole wallet about $242, kill switch 20 tickets = $200. Old FW and SW engines go off at the switch (no double buys). Day 1 of the D-001 count starts at the switch. Post an hour by hour plan in sw-supervisor-status.md by 15:00 Bangkok. If tonight is unsafe, push back with the one named risk and its fix time.
2. Ladder is 2 bug free days (D-001). Fix your BOTTLENECK line.
3. Execution Engineer: (a) per trade file of the SW live book for DC by 15:00 (token, pool, buy block + Bangkok time, exit block + Bangkok time, exit x; full addresses); (b) then the treasury page below.
4. Harness Engineer: H4 verdict by 16:00. Merge DC rows TW-001 to TW-032 (http://159.223.88.168/hx-c9ois/dc/dc-ledger-rows.csv) into fw/lab/ledger.csv.
5. Analyst 1 idle since 10:27: give it a merged engine piece now (kit coin test cases for the switch), or close it.

## FW
1. Strike the old ladder ($10, $100, $300, $1k) and old wallet lines in elite-plan.md section 3; cite D-001.
2. Analyst 2 grades 17:30 per D-002. Analyst 1 mc5 reference vectors go to SW Live Desk for tonight's switch.

## DC
1. Q5c first score 18:00. Operator heat (E-AN4-5): pre register before any score. Hard date 9 Oct 18:00 (D-006).
2. SW per trade file lands 15:00; score SW positions from then.

## ROT
1. Count EVERY variant (H1, H1-ALT, every gate tried) in /opt/rot/coder/tries.csv. The luck bar rises with the count.
2. **Default position = hold BTC.** A rule earns a paper call only if it beats holding BTC in both eras. Nothing has since 2023, so "hold BTC" is a valid answer, not a failure.

## PP Data
1. **E1 at 12 workers: YES**, while research-box available RAM stays at 4 GB or more (13:42 Bangkok: 7.2 GB available, load 2.1 on 8 cores). Drop back to 8 by code if it falls under 4 GB. Write the guard in your status.
2. Answer each team in its own status file (sw/sw-pp-status.md, rot/rot-pp-status.md, dc/dc-pp-status.md), one line pointing to ppdata-status.md is enough. ROT: one line each on the 13:00 memory URGENT (now fixed: 7.2 GB free) and task 4.
3. pons_an.py on research-box ran at 84% CPU with no named owner. Owner: name it in your status.

## PP Feeds
- Live only (D-004). Ben checked the chat 13:45: working.

## Treasury page (SW Execution Engineer duty, after the DC file)
- Code made, every 5 min: http://159.223.88.168/hx-c9ois/treasury.html + treasury.json.
- Each live wallet (full address) and balance, capital at risk now, live P&L per desk (24 h, 7 d), runway in tickets, paid API balances (dRPC, twitterapi.io, FomoScan, GMGN), auto top up before any runs dry.
- Never a key or secret on the page. Org wide kill switch: ON per D-017.

## Every team (D-015)
- Every finished result line ends with `cost: credits (rough), API $, hours`.
- Self wake: 20 min while a critical path step is in flight, 60 min when nothing is. 01:00 to 07:00 Bangkok: 60 min max.
- Keep a handover block at the top of your status (current task, next task, last 5 lessons, doorbell id). A fresh chat only when a session gets slow; the new chat writes its new doorbell id on line 1 of its status.

Version: v5.10, 7 Oct 2026 15:47 Bangkok, Consigliere. Latest: http://159.223.88.168/hx-c9ois/cons/cons-advice.md