IRON LITANY

Published v0.3.2 — find the work

6 October 2026. Runtime frozen and tagged at 77b0e46324b4127f98e7d6ad1e6e83b8fd32e0b1. Later commits retain release evidence and current documentation without rebuilding or changing the shipped game.

Interaction revision

The UX review records the reproduced problems: unclear first-person targeting, terse station labels, insufficient route feedback, competing station actions and unreliable menu focus. The implemented revision adds a current duty card, consistent names, visible walking routes, a YOU marker, written station states, larger held-tool instructions and explicit hands-free/pistol modes. E opens a named reachable station; F changes view; Q draws or holsters; X closes the work face; Escape pauses. The original palette, low-resolution rendering and conditional religious mechanics remain intact.

What was exercised

Ordinary local mouse/keyboard play completed the protected engine rite, restored paid partial work after a save/reload, corrected the Loader, killed two small intruders, sealed one wound and reached the first repair halt. A parts repair and two acts of the conditional cooling rite were exercised; the halt interrupted its last act, so it is not reported as completed. Layout checks at 1280×720, 1024×768 and 390×844 found no horizontal overflow. The narrow check is not a full touch journey.

On the fresh immutable production origin, ordinary controls walked the Engine route, switched to first person, closed with X and reopened the named station with E. Sustained mouse grips completed all three protected acts; releasing the second too early left it unfinished. Its instructions changed with each completed act. Q visibly drew and holstered the pistol; overhead routing and Escape pause worked. No captured console warning/error occurred. The read-only receipt and actual screenshots are retained:

The canonical tab was inspected at its unstarted v0.3.1 welcome, then refreshed to v0.3.2. Continue saved watch remained present. The creator's save was not started or replaced. Owned intermediate QA tabs were closed.

Verification and publication

All 134 game tests passed; JavaScript syntax and 290 local references across 167 shipped files passed. Both native jobs succeeded in CI run 37505271106. Each ZIP passed CRC, full file-manifest checks, pinned runtime digest, x64 executable inspection and exact raw equivalence for all 167 bundled web files. Native startup, view switching, offline pages, saved-origin reload and positive/negative WebGL sight tests passed. CI rendering checks do not establish human desktop combat or hardware performance acceptance.

All six public release assets were streamed anonymously in full and hashed, including both ZIPs and four checksum/manifest sidecars. Linux is 147,209,047 bytes; Windows is 182,276,191 bytes. No duplicate readback files were retained. See [native verification](native-v032/verification.json), [public download readback](native-v032/public-release-verification.json) and download metadata.

The website commit is 1e6145e07f564937586e72419d873069c2783e97. The complete existing production package retained all unrelated game channels, trailers, routes, headers, redirects and edge configuration. All 11 website download/channel tests, source references and public-surface checks passed. Production deployment is 6ac5348afefa47642ef44e7e. Twelve canonical game entry/runtime files matched frozen bytes exactly; the front page matched equivalent parsed HTML after its documented dossier-link rewrite. All three changed website scripts matched prepared bytes. See [canonical readback](native-v032/canonical-web-verification.json) and [publication receipt](native-v032/publication.json).

Owner-private mirror publication terminally succeeded as appgdep_6ac533858f0481919826a1c29680dd35, saved version appgprj_6ac407d6e1b481918ae8e5afe9c4f4ea~appgver_407e35a623708191a7e7e476bcafef95. [Mirror receipt](native-v032/private-mirror.json) records the exact source and unchanged audience.

Limits and resource care

This is an agent-driven expert playtest, not a first-time human usability study. A full three-watch actual-input victory, full touch journey, final physical instrument/priest contact art and sustained human listening remain open. No demo approval or cinematic/narration acceptance is inferred.

Sequential native downloads used a bounded 660 MB peak allowance. Website copying used 100 MB and deployment cache used 500 MB allowances. Both actual resolved filesystems remained above the required 100 GB reserve; final bytes are recorded in the publication receipt. No unrelated files or processes were removed. The dossier was not rebuilt after the runtime freeze.