137 lines
8.0 KiB
Markdown
137 lines
8.0 KiB
Markdown
# Atreides full-plan interaction audit
|
||
|
||
13 September 2026. This audit verifies the existing native Advanced interaction
|
||
paths without promoting an implementation timing choice to an official ruling.
|
||
The public Advanced and unfinished expansion gates remain closed.
|
||
|
||
## Supplied-source No-Field composition — 16 October 2026
|
||
|
||
The authorized supplied revision, physical **p27 C**, restricts the ordinary
|
||
one-element Battle Prescience dial choice against a No-Field. The distinct
|
||
**p27 F** special Karama says to look at any one player's **entire Battle
|
||
Plan** before public revelation. Physical **p36 B** names Prescience in its
|
||
No-Field dial exception; it does not separately name this special power.
|
||
|
||
Under standing choose-recommended, the prototype provisionally keeps those
|
||
powers distinct: ordinary dial Prescience stays prohibited, but native
|
||
Advanced special Karama may inspect the actual complete committed plan.
|
||
This is explicit textual composition, not a newly retrieved combined FAQ or
|
||
a newly answered user question. Other native special-use/counter/mode limits
|
||
and the provisional original target-first offer order remain.
|
||
|
||
The actual target commits first. Only Atreides receives that plan's dial,
|
||
support, leader and physical played faces. This entitlement does not expose
|
||
the token inventory or a prospective maximum to other seats. Private reading
|
||
is not public token revelation: reserves, marker and original event remain
|
||
unchanged until both combatants seal, then the native zero/capped materialization
|
||
and card/cost/loss history run once. This private-read-versus-public-reveal
|
||
composition is provisional as well. No saved field, Basic power or public
|
||
expansion start is added.
|
||
|
||
The same distinction follows native mixed Occupy: an active Ecaz plan uses
|
||
the original fixed increment plus Richese variable dial; prevention restores
|
||
only the chosen lead's own pool. It does not erase the unrelated marker's
|
||
original revelation or alter original card-user funding/physical losses.
|
||
|
||
### Scoped native proof
|
||
|
||
Eighteen new actual-shipment inspection cases and the frozen29-file393/393
|
||
related union pass, with types/lint/build and existing warnings. A final
|
||
type-narrowed real-Traitor case passes1/1. Fixture repairs preserve real shared
|
||
preleader readiness and test an actually legal alternate leader's ordinary
|
||
commitment; no product preparation or target-first guard is weakened.
|
||
|
||
Four actual390px journeys finish original battle-card cleanup/CHOAM market
|
||
into phase7: zero with real leader/cards; mixed ordinary dial3.5/support2;
|
||
Ecaz total5/support0 with marker5 capped to2; Richese H/R total6/support3.
|
||
Each actual private read leaves the native marker, reserves and force arrays
|
||
unchanged. Ordinary Prescience controls still offer only leader/weapon/defense.
|
||
The mixed defeat loses6 real counters. The capped co-side loses4 Richese and3
|
||
Ecaz counters; H/R spends its original3 after materialization, returns2 and
|
||
ends Richese reserves2/Tanks6, Ecaz Tanks3. Native token history remains.
|
||
|
||
Sixteen fresh native programs across all four profiles accept92 policy actions
|
||
through Collection, including original active/prevented owners and actual
|
||
Traitor outcomes. The guide imports with34 evidence paths/source anchor valid;
|
||
five No-Field facets and both topic coverages stay Partial. No overflow/browser/
|
||
application error; owned host/tab/scaffolds removed. Native controls here use
|
||
explicit conserved battle staging after genuine setup/shipment, not complete
|
||
unstaged games, new save assurance or deployed acceptance.
|
||
|
||
|
||
## Published requirements and remaining interpretation
|
||
|
||
The [GF9 base rulebook, printed p14](https://www.gf9games.com/dunegame/wp-content/uploads/Dune-Rulebook.pdf)
|
||
gives Advanced Atreides a once-per-game Karama inspection of any one player's
|
||
entire Battle Plan. Its Q&A p23 orders Voice before ordinary Prescience, permits
|
||
Truthtrance during that interaction and leaves unaffected plan components
|
||
changeable. The Q&A p22 permits a Ghola-revived leader to fight during Battle.
|
||
|
||
The [November 2020 FAQ, pp8–9](https://www.gf9games.com/dune/wp-content/uploads/2020/11/Dune-FAQ-Nov-2020.pdf)
|
||
confirms any-time Truthtrance, current-turn commitments with impossibility release,
|
||
and the Battle-phase Ghola use. The independent source audit retrieved official
|
||
indexed text for these passages.
|
||
|
||
Neither retrieved source orders special full-plan inspection against ordinary
|
||
Voice/Prescience, fixes the special offer's exact deadline, or expressly chooses
|
||
which combatant seals first. This implementation offers it after preparation,
|
||
before either plan is sealed, and requires the selected target to seal first.
|
||
Ghola cannot be played during the pending offer; it can be played after that
|
||
decision resolves. These remain explicit implementation choices. The older WBC
|
||
wording supporting target-first commitment is outside the requested GF9 authority
|
||
and was not adopted as a publisher ruling. Preserve the existing unresolved
|
||
interpretation instead of silently enabling Advanced on the strength of tests.
|
||
|
||
## Verified interaction boundary
|
||
|
||
`tests/full-plan.test.ts` now uses `initializeBaseGameForAudit` and completes actual
|
||
Advanced prediction, traitor and BG advisor setup before staging a battle. It no
|
||
longer starts Basic and flips the mode afterward. A staged scenario remains
|
||
distinct from a setup-to-finish Advanced game.
|
||
|
||
The 17 focused tests include five new interaction tests:
|
||
|
||
- All four profiles can honor a Truthtrance leader/weapon commitment, preserve
|
||
Voice and ordinary Prescience, revive the required leader with one physical
|
||
Ghola, seal for inspection and reach public reveal after the other plan seals.
|
||
- A question during the special offer preserves the pending choice and its
|
||
revival-dependent answers, whether Atreides later uses or declines the special.
|
||
- Questions after inspection disclose only the truthful result of the sealed
|
||
plan. They cannot reopen fields, duplicate a binding record or spend its support.
|
||
- Seven native preparation windows survive a Truthtrance interrupt: Voice
|
||
selection/response, Prescience selection/response/answer, full-plan offer and
|
||
selected-special target wait. Parent response, decision, preparation and plan
|
||
state are retained; named restrictions leave other fields selectable.
|
||
- Atreides can revive its own leader after reading and before sealing, while
|
||
the target's plan remains unchanged and private.
|
||
|
||
Existing tests retain owner-only inspection, sealed-card restrictions, ordinary
|
||
Prescience cancellation, outsiders and once-per-game custody. JSON continuation,
|
||
rejected-action immutability and physical-card conservation are asserted in the
|
||
new paths. No runtime change was needed for these tested interactions. An
|
||
independent review found no actionable issue with the added tests.
|
||
|
||
## Browser and restoration evidence
|
||
|
||
Room `AZC4HGLB` uses genuine Advanced initialization followed by a staged battle.
|
||
An Atreides human chose the special Karama inspection through the browser after
|
||
a Truthtrance leader promise and ordinary Prescience. The Medium Emperor used
|
||
Ghola to revive Hasimir Fenring and sealed a plan honoring those restrictions.
|
||
The private inspection displayed dial 0, support 0, Fenring, Chaumas and Shield,
|
||
while Atreides retained its own plan controls. Expanded component controls and
|
||
the inspection were readable on desktop and at 390×844 phone width.
|
||
|
||
A backed-up restart had already completed when the user removed the hourly
|
||
restart requirement. Browser refresh restored the same owned inspection and
|
||
the earlier inventory room `LC3E55GY`. The final comparison retained all 64
|
||
saved room versions and state hashes with none missing or changed. The server
|
||
returned HTTP 200. Future restarts occur only when a change or observed server
|
||
condition warrants them, with preservation and restoration checks; no recurring
|
||
automation is installed.
|
||
|
||
Required check results are recorded in the checkpoint commit and private
|
||
source-bound report. There are no runtime changes, so the preceding production
|
||
build and HTTP-suite evidence are reused rather than claimed as new runs.
|
||
This audit does not certify arbitrary Truthtrance questions, every Advanced
|
||
interaction, expansion combinations or a complete human/AI game.
|