Improve base guidance and qualify bounded aircraft route and hidden test lifecycle

docs/current-status.md

@@ -1,6 +1,6 @@
 # Current development status
 
-Updated **19 September 2026**. Current addon sources pass the local clean-runtime development gate after fixing native game-over cleanup: [Core](evidence/native-2026-09-19-134325-core-loops/report.json) completes two scripted escape/restart rounds with 612 assertions; [companions](evidence/native-2026-09-19-135014-companion-loops/report.json) complete two with 96. [Ordinary Chat](evidence/game-2026-09-19-135413-native-end-regression/report.json) admits the real canonical prisoner on the same source. Native game-over during preparation and an active prison each pass two rounds with 21 assertions. Those local runs captured no errors. **All seventeen first-release acceptance cases remain pending.** The archived 1317 packages predate this lifecycle correction; their Windows/Linux startup evidence remains tied to those exact artifacts. Packaging UI shutdown errors and four Linux vanilla-asset errors remain unqualified.
+Updated **19 September 2026**. The current 192-file addon snapshot passes [Core](evidence/native-2026-09-19-160849-core-loops/report.json) escape/restart (two rounds, 612 assertions), [companions](evidence/native-2026-09-19-163339-companion-loops/report.json) (two rounds, 96 assertions), and [ordinary Chat](evidence/game-2026-09-19-163745-observer-display/report.json) canonical-prisoner startup with its bounded read-only sampler. Those runs captured no engine errors. The ordinary run also independently verifies the hidden game's window ownership; no keyboard/mouse input or complete player-controlled route is proved. New guidance retries refused native hints, explains rough radio intel and retains preparation/failure feedback. A separate [experimental 20 m aircraft leg](evidence/native-2026-09-19-155909-aircraft-route/report.json) passes 38 assertions including physical translation, landing and shutdown. **All seventeen first-release acceptance cases remain pending.** Existing packages predate these edits and keep their separate qualification limits.
 
 The primary deliverable is an independently authored replacement for the abandoned Escapists base mission, with companion behavior integrated through its interfaces. Five successor addon projects have their own identities, authored Everon sites and vanilla asset references. The first release targets up to twelve players completing prison → scavenging/intel → radio defence → helicopter extraction without the original mod or dependent extensions. Scripted development checkpoints exist; a dependable, manually playable multiplayer release is still unfinished.
 
@@ -16,14 +16,14 @@ The primary deliverable is an independently authored replacement for the abandon
 | Medicine and extraction casualties | [Medical treatment](evidence/native-2026-09-19-062406/report.json) demonstrates finite supplies and native recovery at its earlier snapshot. [Casualty 102037](evidence/native-2026-09-19-102037/report.json) passes 101 clean assertions at the committed checkpoint: grounded-aircraft checks, real incapacitation, helper rejection, seat/LOS constraints, one owner-handshake load and idempotent stop. [Native removal](evidence/native-2026-09-19-080058/report.json) proves the inherited exit mechanism. | Casualty loading explicitly adapts helper membership and invokes action completion. Held input, real remote ownership, competing rescuers and disconnect/death/deletion races remain separate. The older medical snapshot retains its historical catalog errors. |
 | Infantry and HQ | [Actual perception response](evidence/native-2026-09-19-092237/report.json) carries an original soldier's observation through both physical radio legs and HQ to a different original squad, which physically moves. Patrol/preemption/resumption are implemented. | The prisoner is arranged and temporarily protected. Broader combat and interrupted/captured support require additional qualification; no response force is created for this test. |
 | Finite support | [Mortar expiry](evidence/native-2026-09-19-091306/report.json) verifies fresh native commitments followed by stale-report withholding without duplicate orders. [Courier 095513](evidence/native-2026-09-19-095513/report.json) passes 154 clean assertions through actual radio, original-patrol walking, exact depot-shell delivery and patrol restoration. | Mortar observations are synthetic; impact accuracy and genuine late callbacks remain unproved. The new refresh now passes the bounded 112900 fixture: three actual radio updates, eight physical launches, native event completion, 19 surviving original shells and 25 seconds without replay. The current [114751 expiry regression](evidence/native-2026-09-19-114751-mortar-expiry/report.json) also passes 48 clean assertions. [115232 crew registration](evidence/native-2026-09-19-115232-intel-readiness/report.json) proves all four physical crew receive their one Intel carrying roll after setup, then exercises inventory depletion/loss; broader support-loss checks remain separate. Courier depletion is arranged by physically sequestering the original 27 rounds, not firing or replacing them. Broader losses and multiplayer remain unqualified. |
-| Aircraft mechanism | [Flight 114504](evidence/native-2026-09-19-114504-powered-flight/report.json) passes 81 assertions with zero captured errors: signed actor-local collective consumption, native autohover, powered takeoff, five-second 7–9m hover, stable contact landing and an engine-off check with finite fuel. The same aircraft and living AI pilot remain inside the original physical limits. | [Translation 120245](evidence/native-2026-09-19-120245-powered-translation/report.json) additionally passes 85 clean assertions for a fixed four-metre target, five-second hold and landing. The reusable production controller separately passes [123439](evidence/native-2026-09-19-123439-aircraft-controller/report.json), 37 assertions including actual post-spawn qualification and its own complete vertical operation. [Airborne cancellation 124915](evidence/native-2026-09-19-124915-aircraft-cancel/report.json) passes 39 assertions with retained control through contact landing and engine-off. The same reusable controller then passes [fixed-target movement 130038](evidence/native-2026-09-19-130038-aircraft-target/report.json), 38 assertions including a physical 2.9m translation and landing. These are isolated flight mechanisms; rotors still coast after engine shutdown. Longer route navigation, operational HQ aircraft support, repeated-flight reliability and manual/multiplayer piloting remain unqualified. Earlier input, drift, hover and shutdown failures remain archived. |
+| Aircraft mechanism | [Flight 114504](evidence/native-2026-09-19-114504-powered-flight/report.json) passes 81 assertions with zero captured errors: signed actor-local collective consumption, native autohover, powered takeoff, five-second 7–9m hover, stable contact landing and an engine-off check with finite fuel. The same aircraft and living AI pilot remain inside the original physical limits. | [Translation 120245](evidence/native-2026-09-19-120245-powered-translation/report.json) additionally passes 85 clean assertions for a fixed four-metre target, five-second hold and landing. The reusable production controller separately passes [123439](evidence/native-2026-09-19-123439-aircraft-controller/report.json), 37 assertions including actual post-spawn qualification and its own complete vertical operation. [Airborne cancellation 124915](evidence/native-2026-09-19-124915-aircraft-cancel/report.json) passes 39 assertions with retained control through contact landing and engine-off. The same reusable controller then passes [fixed-target movement 130038](evidence/native-2026-09-19-130038-aircraft-target/report.json), 38 assertions including a physical 2.9m translation and landing. These are isolated flight mechanisms; rotors still coast after engine shutdown. [The fixed 20 m route](evidence/native-2026-09-19-155909-aircraft-route/report.json) now passes one unloaded flight, target hold, contact landing and engine-off with 38 assertions. Useful field-length navigation, operational HQ support, repeated-flight reliability and manual/multiplayer piloting remain unqualified. Earlier input, drift, hover and shutdown failures remain archived. |
 | Shared field map | [Map ownership 120633](evidence/native-2026-09-19-120633-map-ownership/report.json) passes native replacement-view, delayed-callback, mode-clear and spare/interrupted deletion checks. The fixture observes actual cleanup and waits for native component references to expire. | Camera restore authorization is observed; no public native render-state getter exists. Visible held-map usability, previously focused stock gadget callbacks and remote presentation remain separate. |
 | Independent runtime controls | [Archived packed Linux Chat 132456](evidence/dedicated-2026-09-19-132456-packed-chat/review.json) compiles the integrated game scripts, loads all five packages and prepares the empty prison mission in an isolated network namespace. Its four resource errors also occur in the earlier vanilla control. All fifteen packed inputs remain unchanged. [Vanilla ordinary 102533](evidence/vanilla-game-2026-09-19-102533/report.json) reproduces the initial splash unknown-class error with exactly two vanilla addons and no successor. | No real dedicated client joined. The vanilla ordinary control remains failed despite entering the menu; it does not test scenario reload or input. [Splash diagnosis](normal-splash-audit.md). |
 
 ## Immediate work
 
 1. **Prove the ordinary player route.** Use normal gate, inventory, document/map, radio and medical interactions; travel the authored route, board and pilot the actual extraction helicopter, and verify partial escape. Keep finite supplies, vanilla recovery and explicit round abandonment intact. Direct-completion fixtures cannot settle these usability questions.
-2. **Extend the proven bounded flight mechanism.** Signed actor-local collective and native autohover now pass vertical takeoff, hover, landing and engine-off checks. A bounded four-metre translation now passes too. Longer routes and operational HQ support still need implementation and native qualification. The reusable controller now passes a native vertical operation; airborne cancellation also passes; optional fixed-target movement now passes too. Longer paths remain the next implementation step. Keep unqualified flight modes out of operational HQ support, retain failed observations, and re-run applicable integration checks after source changes.
+2. **Extend the proven bounded flight mechanism.** Signed actor-local collective and native autohover now pass vertical takeoff, hover, landing and engine-off checks. A bounded four-metre translation and one fixed 20 m route now pass too. Longer routes and operational HQ support still need implementation and native qualification. The reusable controller now passes a native vertical operation; airborne cancellation also passes; optional fixed-target movement now passes too. Operational allocation and useful field-length paths remain next; the 20 m route has only 4.64 s margin in its 160 s translation stage. Keep unqualified flight modes out of operational HQ support, retain failed observations, and re-run applicable integration checks after source changes.
 3. **Establish real multiplayer behavior.** Test accepted client identities, simultaneous/late joins, preload cancellation, reserved-body reconnect/expiry, spent lives, spectator streaming, remote task/map/radio/result updates and seat contention. Local controller injections and helper actors are not substitutes for connected clients.
 4. **Broaden support-loss and observer checks.** Refreshed full salvo, current expiry and late crew registration now have source-bound native evidence; wider support-loss regressions remain. Spectator reason messages and survivor-cycling guidance have been added and compiled; visible behavior and real-client streaming remain unverified. Exercise actual interrupted reports, captured/dead couriers, pending inventory callbacks and support losses. Preserve stock and original patrols; retain historical failures alongside source-bound successful regressions.
 
@@ -33,4 +33,6 @@ The [acceptance ledger](acceptance.json) still requires the complete ordinary es
 
 Collect matched original/successor workloads with repeated client-frame and server-tick measurements. **No FPS improvement is demonstrated**; the [comparison tool](performance-comparison.md) has synthetic regression evidence only. Complete provenance/dependency review, obtain a clean native build and satisfy the [release checker](release-qualification.md). The [packed Core startup](packed-core-startup.md) is now demonstrated for its exact archived PAK, while the [packaging shutdown/control failures](native-packaging-control.md) remain unresolved and unqualified. The [development source bundle](development-source-bundle.md) provides editable projects; it is separate from that native candidate and from a qualified release. No release package is published.
 
+[Private test cleanup](evidence/private-prefix-cleanup-2026-09-19-1631/review.json) now verifies Proton’s trailing-slash prefix spelling, checks idle state under the shared lock, and reaps only positively identified adopted test-device helpers after owned shutdown. The latest companion and ordinary runs record verified cleanup. Earlier strict-byte idle checks could miss those helpers; their historical logs are retained unchanged.
+
 Original missions remain untouched. Engine tests use private profiles/prefixes, hidden displays and muted test audio. RHS, additional terrains, complete original feature parity, alternate starts and distant-unit simulation remain later scope. The [validation ledger](validation.md) preserves the detailed historical checkpoints and failures.