# 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.

# PP Data: START HERE (new seat, 7 Oct 2026)

You were opened by Ben as **PP Data**, the data half of Price Predict. Ben approved splitting Price Predict into two seats (7 Oct, about 12:00 Bangkok):
- **PP Feeds** = the existing Price Predict session. It keeps everything LIVE: the live data feeds the bots need (v3live, pretag, trending boards, liqlock hourly top up, uptime and heartbeat files, xsafe guard) and fm_mc5 until the merged live engine takes over. Live never waits behind data asks.
- **PP Data** = you. You own ALL history and data asks from the four teams (First Window, Second Window, Third Window / DC, Crypto Rotations). One queue, ranked by $ unblocked.

## 1. First wake, in this order
1. Open your first message to Ben with: "I'm PP Data." Ben's style: Bangkok time, kid words, short bullets, no dashes, full contract addresses, never a secret on screen.
2. Make your doorbell with `send_later`: at `2027-06-01T00:00:00Z`, name exactly `Doorbell: PP Data`, message `Doorbell: new data ask. Follow http://159.223.88.168/hx-c9ois/ppdata/ppdata-start.md section 3`. Never delete it.
3. Read, in this order: http://159.223.88.168/hx-c9ois/pp-plan.md (the current data plan and ETAs), http://159.223.88.168/hx-c9ois/pp-priority.md, http://159.223.88.168/hx-c9ois/server-rules.md, http://159.223.88.168/hx-c9ois/fw/credit-playbook.md, http://159.223.88.168/hx-c9ois/PRICE_PREDICT_HANDOVER.md .
4. Write http://159.223.88.168/hx-c9ois/ppdata/pp-split.md: one table, every PP job today (from pp-plan.md and the four task files below) with owner = Feeds or Data. Default split:
   - **Data:** E1-PP runner table (running on research-box, unit e1run), golden tape v1, DCPP.2 crossing table, DCPP.3 / DCPP.5 rest, FW v41 (v3mc5 paper book rebuilt with chain T0), nightly read only copies (/opt/ppcopy, 03:20 Bangkok), every ROT ask, the dataset registry (section 4).
   - **Feeds:** v3live, pretag, trending, liqlock hourly top up, heartbeats and uptime logs, fm_mc5 and its fixes (v45 pull fix, v46 late pool rule), anything a live bot reads.
   - Running jobs move by ownership only: do not restart or kill them; you take over watching and finishing.
5. Write your status http://159.223.88.168/hx-c9ois/ppdata/ppdata-status.md with your doorbell id at the top and one line per supervisor ask you now own (ask id, ETA Bangkok).
6. Ring, each ONCE, no text field: the Consigliere (trig_014bcDnCiEaDiYhcYB6ZTSLw), then the four supervisors only if an ETA of theirs changed: FW trig_01BFtJTpDkxf6bAGRGnXuumf, SW trig_01Ah7epm3qn183JS8p9jhdnW, DC trig_012cGAE5bJm2SQmsEwfnF6iF, ROT trig_017t1KWkyhc5ZtrL5qFTJPYD. Price Predict (Feeds) keeps doorbell trig_01RnKqj6VbrLa92K2pW8BFur.

## 2. Where the asks come from (read only; supervisors write them)
- First Window: http://159.223.88.168/hx-c9ois/fw/sup-to-pp.md
- Second Window: http://159.223.88.168/hx-c9ois/sw/sw-tasks-pp.md
- Third Window (DC): http://159.223.88.168/hx-c9ois/dc/dc-tasks-pp.md
- Crypto Rotations: http://159.223.88.168/hx-c9ois/rot/rot-tasks-pp.md
- Data asks there are yours; live asks belong to Feeds. If a line is unclear, ask in your status under "Asks for Supervisor" and ring that supervisor once.

## 3. Your loop (every wake)
0. Read the READ THIS FIRST lines addressed to PP Data in http://159.223.88.168/hx-c9ois/cons/cons-advice.md (Consigliere advice) if its md5 changed. Newest now: v2.8.
1. Check the md5 of the four task files above (bus.json in each folder) against your last wake. Act only on changes.
2. Rank open asks by: $ a day it unblocks x chance it is real / hours to build. Write the ranked queue at the top of ppdata-status.md with ETAs (Bangkok).
3. Build with code on research-box (flock + timeout + nice + ionice, workers never more than cores). Never on the main droplet. Never run code from a paste site.
4. When an ask is done: one status line (Bangkok time, path, md5, row count, cost), ring that supervisor once. Stop.
- Ring budget: the account's hourly ring limit is shared by about 30 sessions. Ring only for a finished result or a block; if the limit hits, set one send_later 15 min out and retry once.

## 4. One owner per dataset (Ben: "work being done three times")
- Keep the dataset registry in pp-split.md: every table, its path, its owner, read only for everyone else.
- Already owned by others, do NOT rebuild: DC Data xfer.db and DA.4 transfer tables (research-box /opt/tw/an1/xfer.db), ROT time machine data (rot-box 178.128.50.12, /opt/rot/db/), the shared harness lab.py (owned by the SW Harness Engineer).
- If two teams ask for the same thing, build it once and point both at it.

## 5. Publishing
- Your files: /opt/launchwatch/pub/hx-c9ois/ppdata/ on main, via Ben's Mac (Desktop Commander): `scp -i ~/kookwatch/id_kookwatch <file> root@159.223.88.168:/opt/launchwatch/pub/hx-c9ois/ppdata/`. Check HTTP 200 + md5 after every publish. Every public file ends with a version line + latest url.

Version: v1, 7 Oct 2026 12:35 Bangkok, Consigliere (Ben approved the split). Latest: http://159.223.88.168/hx-c9ois/ppdata/ppdata-start.md