8th Wall

You have access to an MCP server that lets you build 8th Wall Desktop AR experiences using Claude Desktop. It exposes a set of tools to create, manage, and run AR scenes, integrate with PolyHaven assets, and drive your projects from natural language. This guide walks you through using the server, installing it, and leveraging its features to build interactive WebAR experiences.

How to use

You use an MCP client to communicate with the 8th Wall MCP Server. Start by ensuring the server is running in either local mode or API mode. In local mode, you run the server on your machine and interact with it directly to create projects, add shapes and models, apply animations, and enable AR features such as image targets or face tracking. In API mode, you can integrate with the 8th Wall Cloud for remote operations. You will issue high-level commands like creating AR experiences from descriptions, adding 3D content, and applying animations, and the server will generate the underlying code and assets for you.

Typical usage patterns include: creating an AR scene from natural language, adding 3D content and physics, configuring AR features, and testing locally with the built-in development server. You can also search PolyHaven for assets, generate JavaScript components, and manage projects from a central console. The system supports both Three.js and AR-specific features like image targets and face tracking, with an emphasis on desktop Three.js workflows for reliable development.

How to install

Prerequisites you need before installation: Node.js (for local runtime), and Claude Desktop installed if you plan to use the one-click flow. You may also want Git if you choose the manual installation path.

Option 1. One-Click Install (Recommended) 1. Download the latest MCP package file. 2. Double-click the file to open with Claude Desktop. 3. Click Install when prompted. 4. Done. All tools are now available for use.

Option 2. Manual Installation 1. Clone the repository from the source location. 2. Install dependencies. 3. Build the project. 4. Add the MCP server to Claude Desktop configuration. 5. Restart Claude Desktop.

If you are configuring the server locally, create or edit your Claude Desktop config to include the standard IO server entry. Use the following JSON structure to run the server with Node.js and the built distribution. The paths must point to your actual dist/index.js location.

{
  "mcpServers": {
    "mcp-8thwall": {
      "command": "node",
      "args": [
        "/absolute/path/to/mcp-8thwall/dist/index.js"
      ]
    }
  }
}

Additional configuration and usage notes

Environment variables can tailor the runtime behavior. You may set modes and paths to your projects. The following example shows a typical environment configuration used when running locally.

{
  "mcpServers": {
    "mcp-8thwall": {
      "command": "node",
      "args": ["/absolute/path/to/mcp-8thwall/dist/index.js"],
      "env": {
        "MODE": "local",
        "PROJECT_ROOT": "/path/to/your/project",
        "EIGHTHWALL_DESKTOP_ROOT": "~/Documents/8th Wall"
      }
    }
  }
}

Security and maintenance

Keep dependencies up to date and follow best practices for local development. If you enable API access, protect your API key and restrict usage to trusted clients. Regularly review environment configurations and verify that local file paths are correct for your project structure.

Known issues & troubleshooting

If you encounter runtime errors, check the environment variables and ensure the dist index is reachable. Some AR features may require testing on a mobile device for camera preview. If an image target camera preview shows black in Desktop mode, verify that the AR setup aligns with Three.js defaults and that the correct engine is active for your project.

Project examples and quick-start commands

These examples demonstrate typical AR scene creation and interaction using the MCP server. Use natural language prompts to describe your scene, then guide the system through content creation, animation, and AR features.

Available tools

create_ar_experience

Build complete AR experiences from natural language descriptions using smart orchestration and code generation.

desktop_add_shape

Add 3D primitives with materials and properties for AR scenes.

desktop_add_model

Insert GLB/GLTF models with positioning, rotation, scale, and animation support.

desktop_add_rotation_animation

Apply rotation animations to objects with axis, speed, looping, and easing.

desktop_add_scale_animation

Create pulsing scale animations with min/max range and timing controls.

desktop_enable_face_tracking

Enable face tracking AR experiences with optional debug mesh.

desktop_add_image_target

Add image target tracking for AR experiences.

desktop_add_video

Attach video textures to surfaces in AR scenes.

desktop_set_model_animation

Control playback of model animation clips with looping and speed control.

assets_search_polyhaven

Search PolyHaven for 3D assets and related media with recommendations.

devserver_start

Start a local development server for previewing scenes.

docs_get_page

Fetch documentation pages for reference.

scene_add_gltf_model

Add GLTF/GLB models to the WebAR scene.

scene_set_background_color

Set the scene background color for visual clarity.

scene_add_light

Add lighting to enhance AR visuals.