ARC-Room: Tactical Mesh Object Intake Lab

May 13, 2026 · View on GitHub

ARC-Room is a lightweight tactical 3D model intake shell for uploading, testing, scaling, commanding, despawning, and exporting objects inside an AI-readable room.

It is built as a separate application from the Stonewalker prototype, but keeps the same tactical mesh / HUD / command-console visual language. The goal is not to replace Blender, Unity, or Unreal. The goal is to provide a fast room-scale object lab that runs locally, loads quickly, and gives humans or AI agents structured spatial state.

Screenshots

ARC-Room intake preview	Stonewalker tactical visual reference

Raw JPEGs are included beside the optimized WebP previews in assets/screenshots/.

What it does

Runs locally by opening index.html.
No install, no build step, no server, no external dependencies.
Uploads .obj and .stl models.
Imports/exports ARC-Room JSON scene files.
Spawns primitive cube and marker test objects.
Selects objects from the list or by clicking the room.
Moves, rotates, scales, duplicates, hides, shows, locks, and despawns objects.
Provides Minecraft-style slash commands.
Shows tactical grid, wireframe objects, bounds, labels, HUD, and command receipts.
Exports AI-readable scene state with transforms, bounds, tags, mesh data, and command history.

Quick start

Open the app directly:

index.html

Or serve it locally:

python3 -m http.server 5173
# then open http://localhost:5173

No package install is required.

Core commands

/help
/list
/select <id>
/summon cube scale=2 at=0,1,0
/summon marker scale=1 at=2,1,-2
/despawn <id>
/despawn selected
/despawn all
/scale <id> 2
/move <id> x y z
/rotate <id> y 90
/duplicate <id>
/tag <id> prop
/hide <id>
/show <id>
/grid on
/grid off
/export json
/save
/load
/reset

Controls

Input	Action
`WASD`	Pan room origin
`Q / E`	Lower / raise room origin
Mouse drag	Orbit camera
Mouse wheel	Zoom
Click object/list item	Select object
`Delete`	Despawn selected object
`F1`	Toggle HUD panels

Supported model formats

Current MVP:

.obj
.stl, ASCII or binary
.json, ARC-Room scene export/import

Planned:

.glb / .gltf
material and texture manifests
object thumbnails
collision/physics metadata
deterministic room replay receipts

AI-readable export

ARC-Room exports structured scene JSON designed for AI/agent pipelines:

{
  "app": "ARC-Room Tactical Mesh Object Intake Lab",
  "version": "0.1.0",
  "room": {
    "units": "meters",
    "gridSize": 1,
    "bounds": [40, 10, 40]
  },
  "objects": [
    {
      "id": "sample_cube_001",
      "name": "sample_cube",
      "position": [0, 1, 0],
      "rotation": [0, 0, 0],
      "scale": [1.2, 1.2, 1.2],
      "tags": ["sample"],
      "visible": true,
      "locked": false,
      "bounds": {
        "min": [-0.6, 0.4, -0.6],
        "max": [0.6, 1.6, 0.6]
      }
    }
  ],
  "receipts": []
}

Repo structure

ARC-Room/
├─ index.html
├─ README.md
├─ LICENSE
├─ CHANGELOG.md
├─ CONTRIBUTING.md
├─ package.json
├─ assets/
│  ├─ screenshots/
│  │  ├─ arc-room-preview.jpeg
│  │  ├─ arc-room-preview.webp
│  │  ├─ stonewalker-reference-boot.jpeg
│  │  └─ stonewalker-reference-boot.webp
│  └─ social/
│     └─ arc-room-card.svg
├─ sample_models/
│  ├─ cube.obj
│  └─ pyramid.obj
└─ docs/
   ├─ COMMANDS.md
   ├─ OBJECT_SCHEMA.md
   ├─ ROADMAP.md
   ├─ RELEASE_CHECKLIST.md
   └─ SCREENSHOTS.md

Why it is intentionally lightweight

ARC-Room is designed as an empty intake shell for testing and developing 3D models on weak hardware. It uses a tactical 2.5D canvas renderer instead of a heavy 3D engine so the first version stays simple, readable, and portable.

The first goal is reliability:

Load model.
Normalize model.
Spawn model.
Transform model.
Export model state.
Repeat.

Physics, animation, multiplayer, networking, accounts, cloud storage, and AI chat integration are intentionally later-stage features.