Preflight worm-ride arrivals for bot candidates

docs/BOT_ARRIVAL_PREFLIGHT.md

@@ -96,3 +96,21 @@ zero rejected candidates and 34 JSON continuations. Both actual
 assassination opportunities were declined; this is not a natural reveal or
 replacement sample. The unsupported reaction combination and public
 expansion gate remain unchanged.
+
+## 29 September: worm ride into Terror with BG intrusion
+
+Another unmodified six-seat Advanced Moritani game at base seed `20261007`
+finished in 1,541 accepted actions, two rejected Fremen worm-ride candidates
+and 41 JSON continuations. Each rejected ride targeted a placed Terror while
+a Bene Gesserit fighter intrusion would open another arrival decision.
+
+The owned bot candidate now quotes the same public destination triggers
+before choosing a worm ride; its current worm decision is the parent, not a
+second reaction. It leaves legal single reactions and supported Ambassador/
+Terror overlap available. The focused regression reproduced the engine
+rejection, then showed all four bot profiles skip the conflicting destination
+and execute an alternative. The same-seed genuine game completes 1,541
+accepted attempts, zero rejects and 41 JSON continuations. No hidden Terror
+face or rival hand enters the quote. Homeworld rides and other special
+arrivals retain their own gates; this does not implement the conflicting
+reaction or open public expansion starts.

docs/CURRENT_STATUS.md

@@ -6,6 +6,14 @@ Expansion starts and full rules/publication acceptance remain gated.**
 
 ## Current checkpoint and work
 
+Another natural six-seat Moritani game exposed two illegal Fremen worm
+rides into a placed Terror with a pending BG fighter intrusion. The
+[worm-arrival bot preflight](BOT_ARRIVAL_PREFLIGHT.md#29-september-worm-ride-into-terror-with-bg-intrusion)
+now omits only that public reaction conflict. Same-seed replay completes
+1,541 accepted actions, zero rejects and 41 JSON restores. This does not
+implement the simultaneous reaction or demonstrate a natural assassination
+reveal; expansion starts and deployment remain gated.
+
 A six-seat Advanced Moritani sample exposed six illegal bot Guild
 cross-planet transport candidates into a placed Terror while Guild income
 was pending. [Arrival preflight](BOT_ARRIVAL_PREFLIGHT.md#29-september-guild-cross-planet-transport-and-terror)

docs/IMPLEMENTATION_STATUS.md

@@ -1,5 +1,23 @@
 # Dune implementation status
 
+## 29 September 2026 — worm-ride arrival candidate correction
+
+A natural six-seat Moritani Advanced sample at base seed `20261007`
+completed 1,541 actions with two rejected Fremen worm rides into a placed
+Terror while a BG intrusion would open. Both were rejected without mutating
+the game. The [owner-view preflight](BOT_ARRIVAL_PREFLIGHT.md#29-september-worm-ride-into-terror-with-bg-intrusion)
+now quotes that public worm destination and excludes the unsupported
+combined reaction before the bot selects an action. A focused engine
+regression reproduced the rejection first and covers four legal AI
+fallbacks. The same genuine game finishes 1,541 accepted attempts, zero
+rejects and 41 JSON continuations. Two assassination opportunities were
+declined; no natural reveal/replacement was proved. The authoritative
+reaction-order gap and public expansion gate remain.
+
+`npm run check` passes typecheck, lint and 5,782/5,782 offline cases.
+No app route or persistence schema changed; build and HTTP checks were not
+repeated for this bot candidate filter.
+
 ## 29 September 2026 — Guild transport arrival preflight
 
 The [arrival preflight](BOT_ARRIVAL_PREFLIGHT.md#29-september-guild-cross-planet-transport-and-terror)