Original Kkrieger: accepted movement input and update progress

The fresh original-game run 20261015-kkrieger-long-input reaches a first-person level after one ordinary 15-second Enter hold, then accepts ordinary W in the original guest input field, advances its update counter twice, and clears both host and guest held state after release. This qualifies key delivery and the identified update-routine progress. It does not measure logical FPS or establish completed simulation, directional displacement, combat, relative mouse, or audio.

Exact source and delivery

Controller 93374 on bx_9ef2wdsw ran source 551caaf97f253928eac65a04cc6e7b91dd260f3e in a fresh clean checkout. All 6521 physical tracked-file SHA256/size records independently match Git blobs; postflight records match preflight exactly. The physically contained 1803537-byte module is 937466db3545b9d016fb90dfd8a414eb82653b5ebbf7cc2db7c53d8cc87d49eb, copied from clean donor 750bc13fd07a59dfed1e49b19533a561ffea1208 only after exact physical build/compiler comparison. An expanded 290-file production/lib/host/ compiler comparison also has no differences. The full intercommit delta is retained; changes concern documentation, tests, and capture/gameplay tools.

The original three files came from /home/user/wa-kkrieger-verified, together with the parent .wine-assembly-browser.json, SHA256 d177f30379d59b6723cba5818bfc37104ef670e9c9549704d1510fd1fd9dba0b. Fresh actual HTTP status/body hashes for module, manifest and all three files passed before launch. The page's own canonical-module digest and loaded-response hash passed, with no WAT fallback. The game has 2147483648 bytes of fixed memory. Viewport remains 1100×800 throughout; there was no Puppeteer observer attachment.

Bounded actions and observations

The controller started 12:51:05.821Z; its absolute capture budget began 12:51:05.908Z with deadline 13:11:05.908Z. Configuration was WA_CAPTURE_MS=1200000 WA_START_HOLD_MS=15000 WA_INPUT_PROGRESS=1. The initial CLI combination --detach --timeout was rejected before execution; the corrected launch uses the controller's internal hard deadline. There was exactly one guest execution.

Scene 48 visibly shows Start new game / credits / exit. The review gate was written 12:59:34.316Z, followed by ordinary Enter held 12:59:36.848–12:59:51.865Z. Scene 62 visibly shows a first-person level with weapon, HUD, lit pillars and floor. Its review gate was written 13:02:27.581Z; the controller consumed it 13:02:35.539Z with over eight minutes remaining, well above the required 160-second collection plus 30-second shutdown reserve.

All four original code-span SHA256 pins passed. The reader kept object 25369268, phase zero, matching focus/main HWND 65537, and pointer lock false. After the 20-second idle control, W was held 13:02:56.137–13:03:24.452Z (28.315 seconds). The guest last-event field became 87 and forward became its positive forward speed 0.002499997615814209, while physical W was true. Subsequent separate samples observed update-counter transitions 390→400→410. The collection reports acceptedProgress: true, two changes. This counter is incremented before the phase branch; it cannot establish completed movement or frames per second.

The first accepted W sample was approximately 8.02 seconds after the first physical-held sample. Idle counter values were 380 then 390; held samples were 390, 400 and 410. The four-code-pin check and original observations are retained in the evidence; there are 10 idle, 15 held and 5 release samples.

After host keyup, the original guest retained its held field briefly. The fifth release sample showed last event 0x80000057, zero forward, and physical W false. releaseObserved: true is independent of the accepted-hold result. The script intentionally keeps gameplayVerified: false.

Personally reviewed idle-before/idle-after and W-before/held/released images show small pillar/lamp/floor shifts. The idle control already changes these landmarks; the held-W shift does not establish distinct directional travel relative to that baseline. Camera settling or bob is possible, not established. Directional displacement remains unqualified; this is not evidence of a key-delivery failure.

The controller exited 0 at 13:03:33.187Z, over seven minutes before its deadline. Postflight at 13:03:50.496Z found unchanged source/module, clean tracked files, and no Node/Chrome/compiler/runtime survivors. Fixture pins also passed in the controller's finally block. The runtime slot was released immediately after postflight. No guest writes, forced transition, forced pointer lock, additional input sequence, deadline extension, runtime patch, or deployment occurred.

Evidence and recovery

Self-contained evidence is under scratch/runs/20261015-kkrieger-long-input/. Remote recovery roots are /home/user/wa-kkrieger-long-input, /home/user/kkrieger-long-input-run, and /home/user/kkrieger-long-receipts; the archive is /home/user/kkrieger-long-final.tar.gz. Recipes in local receipts/ retain preparation, source audit, visual confirmations, download, postflight, and publication steps. Preserve binary evidence independently of Git.

All 16 focused canonical-delivery and progress tests pass with the retained original unpacked image supplied, with no skips. An initial local test invocation lacked the static-server file in the sparse checkout; adding that tracked fixture resolved the setup error. No production code changed in this task.

All 166 remote capture artifacts (24611819 bytes) were copied locally and individually SHA256/size verified. The 24356168-byte archive has SHA256 ca9be1806fe70951c90d5c53ec581e906909c51f20e1ec8d31ef0feb19bbd53e. A transient Boat state: updating rejected the first archive chunk at offset zero; after the box returned healthy/idle the full verified copy succeeded. No resume, recovery, restart, or second guest run was performed. The dashboard result was published last. Its evidence references are all valid; the global checker still reports the unrelated historical Zuma external path.

Root independently reviewed all five control/held/release images and agrees that travel is unqualified. The next source-only question is the original movement consumer and player-position state: distinguish accepted input with small simulation steps, collision, camera motion, or an unconsumed movement field. A future bounded read-only probe needs original-code pins for any new state; this task does not introduce or run that probe.