Milestones

M-ALIGN: Paint = World (playable alignment)
Owner playtest #8 verdict: rooms look PoE2-class but paint-truth and engine-truth diverge — invented painted props with no cells, extent mismatch (painted room bigger than the grid), no reciprocal-door arrival contract, legacy crypt drift, greybox flash at transitions, no walk-behind mask. Bar: every canonical room ≥95% CLEAN on the visual journey; playability-first gate priority ratified.
No due date
•9/13 issues closed
69% complete4 open 9 closed
GA v1.1.0
Act I GA per PRODUCT-ROADMAP §Acts: S-series complete, v1.0.x demo era closed, docs GA-ready (#1346). Collects S10 + GA-blocking items previously scattered in Post-1.0/vNext.
No due date
•0/6 issues closed
0% complete6 open 0 closed
Act II — Harvest Loop (HV-series)
Artifact evals → extract → promote → reuse → flywheel ops. The library grows from scored play; 'less AI dependence' becomes a measured trend. Roadmap §4c.
No due date
•2/5 issues closed
40% complete3 open 2 closed
Act II — Walkable World (W-series)
Scene-at-rest → walk → talk → living stage → the Unity player tier. Exit = the T3 gate: a blind AI playtester completes a quest loop IN the rendered surface. Roadmap §4b.
No due date
•6/10 issues closed
60% complete4 open 6 closed
Graphics ≥8 — Atelier Pipeline
3D modular base (Synty kit) rendered at the contract camera + staging-law light rig (66-80% near-black, 2-4% lit pools) + clean-LoRA albedo-only paint-over + tile-detail finisher. Verdict of the 2026-07-02 strategy deep-dive; supersedes single-image img2img chasing (measured ceiling ~6.0-6.5).
No due date
•1/5 issues closed
20% complete4 open 1 closed
Playable-App Polish
What playtests actually hit: playtest-sourced UX/functional bugs, responsive layout, icon registry, viewer polish, actor/albedo polish (later).
No due date
•3/13 issues closed
23% complete10 open 3 closed
Graphics: PoE2 Playable Combat Demo (Unity 3D-on-2D)
Self-driving roadmap: the repeatable AI system that generates a painterly PoE2-style game world — backdrop (binding >=8) -> pathing -> 3D placeholder actors -> playable combat in-app -> scale. Map: ~/.claude/plans/continue-lets-update-the-wild-lecun.md + docs/roadmap/WORLDOS-GRAPHICS-ROADMAP.md (F3). Tag lane: gfx-* (distinct from the v1.0.x/v1.1 app train). Auto-score every frame to qa/scores_db.py (surface=visual). Baseline 2026-06-30: first combat frame OVERALL 4.08.
No due date
•11/17 issues closed
64% complete6 open 11 closed
The Table (story/world systems)
Post-Beta experience milestone: the features that turn a correct game into one people can't stop playing. Theme: we built the Dungeon Master and the engine; this builds the TABLE around them — the living wait (live composition streaming), visible consequence, companion agendas, session rhythm, fail-forward, map+audio. Specs are implementation-ready for any agent. #1 (Live Composition) is the critical keystone — probe-verified 2026-06-10.
No due date
•0/29 issues closed
0% complete29 open 0 closed
Engine v1.0.5+ (combat epic)
Engine-hardening wave from the 2026-06-11 full-engine audit + post-Beta defects. Pull items into v1.0.4 deliberately, not by default.
No due date
•89/119 issues closed
74% complete30 open 89 closed
WorldOS Web Phase 3 — Fan Showcase + Universe Creator
No due date
•0/4 issues closed
0% complete4 open 0 closed
WorldOS Web Phase 2 — Hosted Runtime
No due date
•0/6 issues closed
0% complete6 open 0 closed
Agent-Grade App Testability — Built-App Harness & Evidence
Make dist/WorldOS.app observable and driveable by agents without replacing the real release gate: deterministic smoke provider, app-status v1, stable UI hooks, evidence export, and split gates. Engine remains sole writer; full RRI remains the release verdict.
No due date
•16/16 issues closed
100% complete0 open 16 closed
Post-1.0 / vNext
Parked someday/roadmap work: universes (#330-332), multiplayer (#31), distribution (#32), optional renderers, and the strategy epics (#643-645). Not in an active sprint.
No due date
•1/16 issues closed
6% complete15 open 1 closed