Adaptive XR Difficulty-Adjustment Framework — Pilot Data and Code (Engineering Feasibility Study)

Complete data and code archive for the manuscript "An Adaptive XR Framework for Real-Time Difficulty Personalization: A Pilot Engineering Feasibility Study" (submitted to IEEE Access, manuscript Access-2026-28737). The archive contains: (1) all 16 session logs (CSV) from the pilot campaign — 12 anonymized participant sessions and 4 empty logs from aborted technical test runs (282 interaction events; 86 release trials; no records excluded); (2) the historical C# scripts recovered from the archived Unity project (Unity 2022.3.62f3 LTS, XR Interaction Toolkit 2.6.5, Meta Quest 2), including the proportional difficulty controller; (3) retained-scene configuration evidence: the deployed scene file (MainScene.unity) with its asset GUIDs, Unity/package version records, and a scene-derived machine-readable configuration index, plus the SHA-256 of the retained pilot build; (4) a verifying Python analysis script that checks the archive against a cryptographic manifest (SHA256SUMS), validates the structure of every session log, re-derives the controller configuration from the archived scene, replays the controller rule (reproducing all 86 logged trial difficulties at the recorded precision), recomputes the reported correlations, and regenerates the data-derived tables and figures of the paper from the raw logs; (5) a README documenting all column definitions, logging conventions, controller parameters, release-position handling, and the archive's known limitations. The study is an engineering feasibility evaluation with non-clinical adult volunteers. It makes no claims about learning, engagement, or clinical effectiveness, and involved no autistic participants. No personal, demographic, or health information was collected; sessions carry pseudonymous three-digit codes derived from the session start time, and the deposited files contain no direct identifiers or participant-key mapping. This is a pilot analysis and historical-source archive, not a complete experimental-replication package: the full Unity project and build are not redistributed, and no per-session build identifier links individual sessions to a specific build. Version 1.1 corrects the README: manuscript title updated; release-position handling documented as varying across pilot builds (earlier sessions log exact-zero errors); reproducibility scope clarified; replay verification added (replaying the recovered controller rule over the logged outcomes reproduces all 86 logged trial difficulties exactly); note added that reaction time is logged but not used by the controller. Version 1.2 adds the retained-scene configuration evidence described above and hardens the analysis script into a full verifier; output filenames are aligned to the manuscript's table numbering; a CHANGELOG and full-coverage SHA-256 manifest are included. Raw data and historical scripts are byte-identical to v1.0/v1.1. Release ZIP SHA-256: 2cf90dfafb00194ca797830b7df5ef1bd51d5a6c0831e5cc91eed7fa595caaaf Code: MIT License. Data (data/*.csv): CC BY 4.0.