Penguin Command: original A shot accepted, 2026-10-24

The captured original-game shot is accepted: active shot count at 4130a4 changes 0→1, and effective cannon ammo at 410110 changes 10→9. This is a historical headless observation, not a current-main Wasm qualification. Visible projectile appearance remains unproven: optional PNG writes at batches 39999 and 40100 both failed with EFBIG and produced incomplete files. No score, wave, or general playability conclusion follows.

The complete raw log contains the decisive block entries 406ec4 → 406ed8 → 406efb → 406f00. At the first entry, EDI=1 and ECX=0, count is0 and ammo is10/10/10. The original PE instructions bind 406ecb lea eax,[ecx+1] and 406ece mov [4130a4],eax to count1 at 406ed8. 406ec4 mov edx,[41010c+edi*4] selects 410110; EDX=10 and lazy comparison operands10−10=0 at 406ed8 confirm the ammo comparison fell through. 406ee4 lea eax,[edx-1], 406ee7 mov edx,410110 and 406eec mov [edx+edi*4-4],eax bind the decrement to that same slot. The post-store call block 406efb and epilogue entry 406f00 retain count1 and ammo9/10/10. Fall-through instructions need not appear as separate block-entry lines.

The first four entries retain shot record0 at 4d0750 initialized to floats50/515 and velocity bytes; reverse remains0 and target coordinates237/177. Two later entries at 406f40 belong to the next routine. Their later ammo−1 and changed shot coordinates remain in the raw log; the immediate-store finding does not extend to those later effects, and no defect is inferred from them.

The same log records actual A delivery: keydown msg100 wParam41 lParam1e0001 through focus window10001 at batch40033, then keyup msg101 wParam41 lParamc01e0001 at40295. Scheduled batches were40000/40200. Existing SDL/original-filter attribution is scoped predecessor evidence in 85258092, d8646d93 and 164e406b.

Exact argv uses the existing runner with no build, range 406ec4-406f50 excluding the4000-iteration initialization loop, stream/detail memory tracing, menu34000/34500, aim34800 and balanced A40000/40200. No force-state option or guest modification was used. Supervisor source passes argv directly and clears the extra-argument and region-map environment overrides. Output is13171 bytes, six EIP entries, final newline and normal final statistics at43000 batches, below the524288-byte guard. Terminal receipt and launch receipt agree on natural return0 after3.154seconds, without termination, kill or remaining process group.

Fresh identity evidence, the 291-file manifest and remote rehash receipt bind runner SHA256 3c59fd034391e069f24a6047cf96cd8dac84a791d7d7591d55e5f7cb6e161a5e to main5691793a, original EXE 14d6e430c26c44c94ef5c176eca2f66edd82dd66fd155d29a5596e3d6e170fc1, and historical Wasm 8fac39936b6df2418ad86cab118cef19a65a727f649fbc14c412b37818904686 from b41a7c5f. That module is not current-main Wasm. The manifest's original absolute donor paths describe provenance; reviewable proof is retained here or through the original Git commit below.

Independent review and rejecting probes accept this limited conclusion. The original task tree retains all33 sealed artifacts, including the failed PNGs and API history. Those PNGs are failed artifacts, not usable frames. The original full task report preserves capture/resource history. Its RUNNING lifecycle statement was historical: the coordinator subsequently verified bx_fq6tc9p2 snapshot STOPPED at06:13:07.152Z; this independent review performed no lifecycle operation.