Native control, inspection, UI automation and evidence capture for a running Witcher 3 process.

mod witcher3

Fork 0

witcher3-devkit docs QUALIFICATION.md

2.2 kB · 36 lines

Markdown

at main

iostreamer.at · Record exact updated-game qualification and recovery limits · 2w ago

7355513e

Qualification boundary #

Steam build 26238013, executable version 5.0.16.41566, subsequently passed 16 bounded native compatibility checks with the patched WNB resolver, plus 384 offline tests in both worktree and source-only environments. The executable was not modified or rolled back. The exact hashes, failed shader-delay startup, successful retry and restoration evidence are in the reviewed update report. See the reconciliation contract before accepting any update. This is not an OBS/audio capture qualification or coverage of every native API.

The instrumented game was checked on Remastered 5.01 DX12 under Steam/Proton and KDE KWin. The organized source passed 318 offline tests in the worktree and an isolated source-only archive. Build a6efd7dea288939c then passed 31 bounded cold-start debugger checks in 129.81 seconds at source a8197fd: hashed load, bounded input, sampling, owner reconnect/handoff/recovery, Rust export, unsaved close and exact prelaunch save/mod/display-settings restoration. The separately retained native session's input-settings hash also matched after close. The private report is debug-acceptance-1792934590914205296/report.json under the configured installation's artifacts directory (SHA-256 dcd05c01cd8d1453a19102016a699749c078eff96fbf44505b73b4853b820b93). The ordered WitcherScript concatenation and installed bridge build are byte-identical to the previous release. A rebuilt HUD preserved 107 other UI resources and reproduced both installed owned-asset hashes.

The generic selected-actor HUD list previously passed 19 bounded 1920×1080 live checks with pixel review. Earlier header-only visual failures were retained in the private validation history. A separate consumer confirmed an actual eight-entry memory list, but its dense fight window had an event gap; that is not full-fight evidence.

Source-only tests and byte-identical native outputs do not qualify an arbitrary game installation or a newly built runtime. There is no claim of measured physical input latency, deterministic replay, complete AI observability, cross-resolution layout, or general first-time installation.