Dune/docs/SPICE_BANKER_RUNTIME.md

Spice Banker battle commitment and normal income prototypes

2 October normal-band checkpoint: a separate fresh local banker-income classic/base33/all-fourteen-skills-only Basic/Advanced profile connects actual positive bank payments to automatic current Mentat collection. Bounded coverage is Partial / Development Verified. No public start, old-game retrofit or other module is enabled; this is not deployed acceptance.

Normal income local contract — 2 October 2026

The original photographed card says gain one spice when another player makes one bank payment of at least four, once per phase. This is a mandatory additional bank grant, not skimming the original payer or recipient; no new action, Karama response or confirmation is invented. Ordinary costs and their original counters/recipient/physical suffix finish once before the grant.

The private-backed-up profile admits unique2..6 classic factions, Basic or Advanced, base33 Treachery and all14 Leader Skills alone. Actual paid normal auctions, funded shipments, paid force/leader/KH/Emperor-extra revivals and resolved bank support use original final-bank receipts. Player income, free/waived/stopped/invalid sources and self-payment do not qualify. The original bank recipient matters: canceled Emperor or Guild income may leave a bank payment, while income actually received by a player does not.

One payment is not several smaller payments summed together. Actual own/donor legs keep their payer identity; two payers at two each do not make one payer at four. The lower1..3 Banker battle payment stays separate from support. This per-payer/source composition is a conservative local interpretation, not a separately located publisher split-funding example. No future award can finance the original payment.

An actual available living native trainer supplies new normal income. A captive supplies no upper band; concealed unselected skills are unavailable. Own-battle selected-trainer eligibility uses actual resolved survival, not private pre-reveal timing. One physical card's turn/phase stamp cannot reset through death, return, refresh or same-phase reassignment.

The already-earned front-shield pile remains bound to its ORIGINAL gaining faction provisionally in this fresh profile even if the trainer dies, is captured or the card returns/reassigns. This conserves earned currency and does not transfer it with a skill card. It does not answer the previously asked death/capture question or create universal completed-mode authority. The historical pending boundary below remains recorded outside this profile.

Deferred spice is not added to a spendable wallet before actual native Mentat opening. Human/four-policy bids, shipping, revival, support and lower Banker spending use original available money. Mentat atomically marks/credits each uncollected grant once before optional work/victory; collection is not a qualifying payment and creates no recursive income. There is no arbitrary read heal or fictitious persisted half-phase collect choice.

Only physical public front-shield owner/count and current-phase use are projected. Private bank amounts, source/receipt, hidden plans and later skill entitlement stay private. Current-source corrupt/stale/duplicate/foreign actions and recovery reject immutably; closed history no longer requires the original trainer/card or old wallet to remain in place. Bounded signatures check consistency, not authentication of wholesale rewritten saved JSON.

Frozen verification passes types/lint, 6,487 offline tests, the 140-case Banker/profile union and build. Eighteen actual native producer/threshold/ trainer cases conserve all33 Treachery Cards and14 skills through original payment and real Mentat; independent combat/Collection quotes separate native bounty/income from support costs. All four minimal legal policies exclude the front-shield pile from available budgets. Native death/capture/reassignment, once-phase and separate payer legs have meaningful JSON/authenticated SQLite regressions, including actual final Collection races and private suspended Emperor/Guild invoices. Two final independent scoped reviews have no finding.

Original private-backed-up six-seat CLI setups were continued without redealing: Basic 6VZ8G5BY v38 used native Emperor self-auction payment to the Bank; Advanced 8TN8BTLT v31 used native Guild own-bank shipment and its genuine allow decision. Human390px controls earned1 without changing the Banker's available wallet, retained the notice through refresh and collected at actual final Collection readiness to Mentat exactly once (10→11 and3→4). The notice's initial26px player-strip squeeze was reproduced visually and fixed by placing it in the main column; fresh proof shows354px width. All cards/skills/normal and elite force inventories remain conserved, with the23 prior isolated QA rooms' states/versions unchanged. Final guide10/10, types/lint/build and expanded390px guidance pass; all82 original seats in25 targeted QA rooms restore after final worker reload without double collection. Local HTTP is 49/55, with six known upstream503 rejection failures: not a green suite. Publication/deployment and complete skill/module acceptance remain open.

The historical13 September checkpoint below is Prototyped, with Partial rules coverage, and connects the lower battle band only. The normal band is now connected solely in the fresh profile above; combined acceptance and public mode starts remain gated.

Source and payment contract

The physical card permits spending one through three own spice with the Battle Plan to add the same amount to the selected leader's strength, independently of force support. A captured skilled leader can supply this battle band with the captor's spice. The common skill rules remove the bonus if that leader dies. See GF9 CHOAM & Richese rules and Q&A, pages 8–9 and 12, and Jack Reda's walkthrough, 17:09–17:37. The existing physical card image and sourced full paraphrase are retained; this creates no new card, resource or leader identity.

The core rulebook provides the payment precedence in its Traitor and Advanced Combat sections: a sole successful traitor caller loses nothing; otherwise committed Battle Plan spice goes to the Bank on either outcome. The prototype therefore spends Banker spice even on leader death or traitor loss, and waives it for a sole traitor-call winner. Death removes the strength bonus, not the already chosen plan cost. This is an explicit composition of the card and core payment rules, not a separately located Banker death FAQ. It supersedes the earlier private research suggestion that death should release the commitment. Printed disc strength, Zoal's copied value and leader bounty are unchanged.

The normal band is mandatory once per phase when another player makes one payment of at least four spice to the Bank; multiple smaller payments do not combine. Revival qualifies. Earned spice waits in front of the shield until Mentat Pause. Neither publisher nor designer material located in the renewed search settles custody of that accumulated income if the trainer dies or is captured first. The user has been asked whether to retain the pending boundary or rule that already-earned spice stays with the original faction. No answer is assumed; only the explicitly provisional fresh profile above implements normal income. This historical question remains open outside that profile.

Connected behavior

bankerSpice is an optional sealed-plan amount; absence or zero declines. The engine validates an actual living controlled skilled disc, its battle posture, integer range and own funds remaining after force support. Ally aid and Stronghold support cannot fund this separate commitment. Saved plans reserve the amount against bribes and other spending. The same checks run before views, automatic normalization and actions. Legacy plans retain zero Banker spending.

The battle quote computes a separate payment alongside the existing support settlement and a surviving-leader score bonus. It does not add Banker spice to CHOAM's force-support share. Payment and resolution commit atomically through the existing versioned room write, with an automatic explanatory history entry.

The battle panel distinguishes support and Banker amounts and shows the combined own commitment. The owner and authorized full-plan inspector can see the sealed amount; rivals see it only when both plans are public. All four AI profiles use the public skill and private own budget to submit funded plans. This is legal prototype behavior, not calibrated Banker strategy.

Evidence and remaining scope

Focused engine checks cover one through three spice, win/loss, death, traitor win/loss/mutual calls, captured-leader funds, printed values, own-support funding, private views, malformed saved fields, JSON restoration and all four profiles. SQLite checks cover reserved-spice diversion, private restart, concurrent final resolution, exact waiver/bounty and legacy absence. Controls and inspector checks retain the private/public boundary. Independent rules, persistence and UI review cover this supported boundary.

Two genuine four-profile Basic/Advanced samples completed with 99/212 actions, 3/6 JSON restorations and no rejected candidates. Their first seat explicitly selected its offered Banker, but neither game used Banker spending. These are continuation samples; the targeted tests establish the effect. A fresh browser room assigned Banker through genuine setup, followed by a conserved staged battle. The player selected Hasimir Fenring, Baliset, four dialed forces and three Banker spice. Pending refresh restored that commitment with five spice still held. Settlement paid three, added three strength and left one Arrakeen force, fifteen reserves, four Tanks and two spice. Another refresh retained those values and the same private Crysknife, Baliset and Burseg while card cleanup awaited the player. This targeted scenario is not a complete human game. Required broad results are recorded with the checkpoint.

Remaining work includes normal income, all economic trigger routes and ordering, its public deferred custody, combined modules, wider full-game interaction and strategy refinement. The existing development gates remain unchanged.