Researcher (Haiku) scanned all 6 Ventilatore packs (Premium, Castles
and Fortresses, Medieval Interiors, Desert Oasis, Snow Adventures,
Turning of the Seasons). The bundle is character + terrain + decoration
only — no animal/creature sprites at all except player + slimes.
Wolf options now narrowed to: commission, CC0 source, or recolor-a-dog
placeholder until Phase 10. Ventilatore-search sub-option closed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wood walls (early game) custom-authored on top of FG_Houses warm-brown
timber palette to preserve the Stardew-cabin warmth that's the project's
aesthetic anchor. ~½ day of pixel art for corner/T/cap/cross variants.
Stone walls (upgrade material) imported from FG_Fortress autotile-solvable
as-is. Both materials plug into the same WallMaterial enum / construction
pipeline. Phase 5 estimate bumped from 2-3 wks to 2.5-3.5 wks.
Phase 1 wall-rendering test now specified: use FG_Fortress as the
drop-in test material; wood walls land in Phase 5 alongside the
authoring task.
Iconic Homestead $19.99 fallback formally not needed.
memory.md decisions table now includes the wall locks; the original
3-option open question collapsed to a back-reference. art.md got a
new 'Wall-material decision' section replacing the options block.
Wolf-sprite question still open; Ventilatore search dispatched.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Project scaffold:
- project.godot at repo root, GL Compatibility renderer (max mobile reach),
pixel-snap on, texture filter nearest, sensor_landscape orientation
- 7 autoloads: World, Sim, GameState, EventBus, Strings, Audit, SaveSystem
- scenes/main/main.{tscn,gd} smoke-test scene with autoload assertions
- Folder layout matches tavernkeep idiom: autoload/ at root, scripts
co-located with scenes/ (not the scripts/autoloads/ mirror originally
sketched in implementation.md)
- Input map: pause, speed_cycle, speed_normal/fast/ultra, confirm, cancel.
Mobile gestures (pinch/drag/long-press) handled at script level via
Godot's InputEventScreenTouch/Drag/MagnifyGesture.
- SaveSystem skeleton: SAVE_VERSION=1, JSON to user://save_slot.json,
version-mismatch warning. Phase 3 expands to real entity state.
- icon.svg placeholder (cabin silhouette on dark green field)
- README.md points at memory.md / implementation.md / docs/
Headless verification: 'godot --headless --path . --quit' exits 0,
'[main] Phase 0 smoke test online.' prints, no errors. Editor-side
green-dot check still pending — needs human launch of editor.
Asset audit (researcher Haiku, 2026-05-10):
- FG_Houses.png NOT autotile-solvable — pre-built decorative house
compositions, 4 distinct roof palettes, no modular wall family.
~½–1 day per material to author terrain bits on top.
- FG_Fortress.png IS autotile-solvable — ~20–30 modular tan-stone
pieces. Wang-style Godot 4 terrain works with minimal extra art.
Iconic Homestead $19.99 fallback not needed.
- No wolf sprite anywhere in the bundle. EvoMonster packs all
cute/fantasy. Need commission, CC0 source, or Ventilatore check.
- Retro Graveyard 16x16 [Kingdom Explorer] confirmed in Tier 3 with
full graveyard suite — direct use in Phase 14.
New open questions surfaced in memory.md:
- Player-built wall material strategy (3 options laid out)
- Wolf sprite acquisition path (Phase 10 blocker)
Project move:
- Repo physical location moved from ~/claude/projects/rimlike to
/mnt/d/godot/rimlike (D: drive, fast for Windows-side editor).
- Symlink at the original WSL path preserves the home-CLAUDE.md
layout convention. Mirrors tavernkeep's pattern.
- Set core.filemode=false to silence DrvFs's everything-is-0777
false-positive on git diff.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
21 phases from clean-slate to MVP, dependency-ordered with each
phase ending in a runnable demo state. Includes:
- Pre-implementation audit (the 75-min asset checks from memory.md)
- Phases 0–20 with checklist + acceptance criteria + spec-doc refs
- Two DECIDE points (Phase 13 big-room UX, Phase 19 onboarding)
- Out-of-scope list (procgen, biomes, butchering, surgery, etc.)
- Scope-cut levers ranked by gameplay-cost-per-week-saved
- De-risking spikes per phase
- Wired into memory.md index + session log; current Status =
Phase 0 (gated by the 75-min audit)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Resolves the rimlike docs audit. Decisions made this session:
- Map: 40² → 80² for MVP slice; architecture sized to ~120² ceiling.
Pawn count unchanged (3 start / 6 cap) — 'split-the-difference'
sizing rather than frontier-feel scale.
- World-view camera (locked): pinch + drag-pan + double-tap-centre.
Storyteller alerts include 'Go there' tap. No minimap, no
follow-cam in MVP. New ui.md section captures this.
- Storyteller cooldowns: per-event AND per-category — both gates
must pass. Resolves design.md/architecture.md disagreement.
- MAX_STACKS_PER_THOUGHT = 5 (was named-but-unset).
- Hauling no-destination fallback: drop after 3 retry passes,
surface a passive 'No stockpile accepts X' alert.
- Container all-neighbors-blocked: hold then drop after ~5 sim s
to avoid deadlock.
- Auto-roof big-room UX: emits room_too_large signal when BFS hits
the ≤8-cell cap; UI surfaces a 'split with an interior wall'
banner. Threshold + exact wording added to memory.md TODOs.
Doc rot cleaned in design.md and architecture.md: removed the stale
'8 work categories' intermediate sections (canonical 9-list lives
later in each file). Updated architecture.md perf claims for the new
map size.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Promoted from ~/claude/ideas/rimlike after a single multi-hour brainstorm
session. memory.md distilled from plan.md; companion design / architecture /
ui / art docs preserved under docs/.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
