Original Penguin Command 1.6.11 execution, 2026-10-24

Historical source last updated, 2026-10-24. Runtime and implementation claims below describe that dated source; see current app notes and source for subsequent changes.

BLOCKED by the executed output guard; ordinary gameplay is unverified. The real original EXE started, all seven native providers loaded and returned success from DllMain, and an SDL window acquired the title “Penguin Command”. The inspected 640×480 exit frame has a black client area. It establishes neither the menu nor playability. No ordinary input was sent, and no menu, aim, fire, projectile, enemy, score or wave response was observed.

Executed blocker

At 2026-10-24T02:14:00.383Z, API command cmd_rstynxdxceqw launched the bounded supervisor on exclusively assigned bx_9nngcsvf. The guest log grew past the supervisor’s 524,288-byte captured-output threshold because this historical runner prints per-batch EIP/register progress by default. The observed retained file is 554,504 bytes, including orderly shutdown output; the polling guard is not a byte-exact truncating writer. The supervisor sent TERM and recorded exit 0, elapsed 1.26752 seconds, no KILL, no additional reaped children and no remaining process group. The runner reported 761 API calls and 4,666 batches over 0.721 seconds. Exit 0 is not an acceptance result.

The first scheduled control snapshot/capture was at four seconds and was never sent: run/input-events.json is the actual empty list. No genuine guest trap, unimplemented API, import failure or guest asset failure was observed in the retained log. Static WINMM warnings were not treated as failure. Execution stopped at the guard rejection; there was no retry or speculative guest repair.

Exact inputs and preflight

The original EXE is 107,008 bytes, SHA-256 14d6e430c26c44c94ef5c176eca2f66edd82dd66fd155d29a5596e3d6e170fc1; the wrapper was not used. Standing MSVCRT is 278,581 bytes, SHA-256 887eb5ce93edb7192ca3e9220f07f9ca0f94db02af5862ebcbdfcb852db99fd1. All seven explicit providers and all 83 original PNG/JPG graphics are bound by path, byte length and full SHA-256 in adjacent admission.json. The ordinary flags were --windowed --nosound, guest EXE C:\penguin-command.exe, CWD C:\. SDL_mixer remained loaded. Original closure acceptance 98257ee06ff0e68900025f20ab1453379e34b6ed was reused without another archive/dependency audit.

The retained pngjs 7.0.0 package came from /home/user/wine-assembly/scratch/usage-watch/shadow-man-codec-recovery-oct22/node_modules/pngjs. All 23 receipt-bound files, 63,864 bytes, were rehashed and staged unchanged at runner/node_modules/pngjs; receipt paths have their leading runner/ stripped when reading the standing retention. Both the local 2×2 and remote 1×1 PNG encode/decode controls passed pixel equality. Remote ordinary module resolution and production rasterCanvas/renderer loading passed. No skia decodeImage profile or headlessGL route was selected; original game decoding uses its bundled DLLs.

All 125 supplied historical manifest files were read through exact git show with GIT_NO_LAZY_FETCH=1 and matched their sealed hashes. Two omissions in that supplied manifest were corrected from the same canonical Git donor:

This was a routine runtime-data manifest correction, with no harness or guest source change. The superseded first preflight receipt remains as preflight-incomplete-font-data.json. The corrected preflight explicitly loads the renderer and reads every derived font mount. It verifies 287 files, 15,719,953 bytes, all 83 game graphics and the exact full module hash. The accepted full-module ABI contract is reused by that exact hash, including 291 imports and shared memory 8192..32768 pages. The inert module layout stamp and exact historical JS mirror both match 9a3bc9149d58209c. Actual native instantiation and DLL initialization subsequently succeeded in the game run; this is not an independent all-signatures ABI reacceptance.

Recovery and evidence

Adjacent preparation, transfer, correction, preflight and supervisor scripts preserve the executed recipe and all input hashes. They have fixed one-task destinations. executed-supervise.py.gz retains the byte-exact supervisor; the readable .py copy only removes trailing whitespace, with both identities in supervisor-formatting.json. The original collection seal binds the archive’s decompressed bytes. Do not replay these scripts over retained evidence. Add --quiet-blocks to the recorded argv without relaxing guards; then inspect menu/game frames and send ordinary inputs only if actually reached. No new execution was performed to test that proposal.

There is no Penguin candidate in this exact main’s candidate corpus, so these captures stay with the investigation and are not attached to another app. results.json is published last as investigation metadata, not a dashboard candidate/gameplay acceptance. validation.json records local syntax, retained hash and evidence checks. No runtime/gameplay success is inferred from those host-only checks.