HomeGuidesChangelogDiscussions
Hire a ReactVision Expert 🛠️Launch Studio 🖥️Sponsor on GitHub ❤️Log In
Guides

Overview

An introduction to ReactVision and our XR development tools.

ReactVision is the XR platform AI agents build with. You build in React Native, so your app is code your agent already understands, and the learning curve for you is short.

Write once, ship everywhere: AR on iOS and Android phones, mixed reality and VR on Meta Quest, and fully immersive apps on Apple Vision Pro.

Use it either way: build a new AR or VR experience from scratch, or add XR features to an app you already ship. Build for one person or for several in the same room: co-located sessions, shared state and the coordinate maths behind them are part of the renderer, not something you bolt on.

ReactVision's XR stack is made up of four parts:

  • ViroReact is the renderer, open source under the MIT licence. It draws your scene natively through ARKit on iOS, ARCore on Android, OpenXR on Meta Quest and Metal on visionOS. You build with React components (<ViroARScene>, <Viro3DObject>, <ViroImage>) and ViroReact handles the native side.
  • Platform is the managed backend for the features that need infrastructure: Cloud Anchors, Geospatial Anchors, co-location rooms and the replication relay, asset storage and hosting for over the air 3D content updates, and environment testing.
  • Studio is a CMS for XR, in the browser. Compose scenes visually, generate assets with AI, preview instantly, then drop the result into a React Native app with StudioSceneNavigator. The Studio Agent builds scenes from a description.
  • MCP Server connects your coding agent to all of it. Point Claude, Cursor or Windsurf at https://mcp.reactvision.xyz/viro for current API knowledge, scene validation, asset generation and real renderer testing.

📘

The renderer is open source

ViroReact has been MIT licensed since v2.17.0. The source is on GitHub and it is actively maintained by ReactVision and the community. Platform, Studio and the MCP Server are the hosted layers on top.

What You Can Build

  • AR on phones. Plane detection, image and object tracking, portals, depth and occlusion, and world scale anchoring through Cloud Anchors and Geospatial Anchors.
  • Mixed reality and VR on Quest. Passthrough scenes, hand and controller input, eye gaze on Quest Pro, and room geometry from Space Setup.
  • Fully immersive apps on Apple Vision Pro. ImmersiveSpace scenes with PBR materials, particles, physics, custom shaders and skeletal animation.
  • Shared, multi-user experiences. <ViroARCloudAnchor> puts two or more devices in one agreed coordinate frame, so a child at [0, 0, -1] is the same physical metre on every device. useViroReplicatedState keeps the room's state ordered and conflict resolved, with objects claimed before they are changed, so two people grabbing the same thing resolves rather than races. Smoothing hooks keep peer and object motion fluid at any frame rate.
  • Scene understanding. Scene semantics, world mesh capture, on device object detection with ViroObjectDetector, and AR session recording for offline analysis.

Supported Platforms

🍎 iOS (ARKit)🤖 Android (ARCore)🎮 Meta Quest (OpenXR)🥽 visionOS🤖 Android XR
StatusGAGAGAGAComing soon
Plane detection (horizontal and vertical)✅✅✅––
Image tracking✅✅–––
Object tracking✅––––
Passthrough mixed reality––✅––
Immersive space––✅✅–
Hand and eye tracking––✅––
Cloud Anchors✅✅–––
Geospatial anchoring✅✅–––
Co-located sessions (shared frame + state)✅✅–✅–
Scene semantics–✅–––
Object detection (ViroObjectDetector)✅✅✅––
AR session recording✅✅–––
Depth and occlusion✅––––
Portals✅✅–––
PBR, particles, 360° images✅✅✅✅–
360° video✅✅✅––
Spatial audio✅✅✅––
Custom shaders✅✅–✅–
Skeletal animation✅✅✅✅–
Morph targets✅✅–✅–
Physics engine (Bullet3)✅✅–✅–
Studio scenes✅✅✅––

Quest support covers Quest 2, Quest 3, Quest 3S and Quest Pro. Plane detection, passthrough and mixed reality scenes need a Quest 3 or 3S, object detection needs Horizon OS v74 or later, and eye gaze needs a Quest Pro. On Quest, Studio scenes render through ViroXRSceneNavigator. visionOS covers fully immersive apps on Apple Vision Pro rather than passthrough AR, and loads GLB and glTF models. On iOS, depth and occlusion need a LiDAR equipped device, and geospatial runs on VPS-Lite rather than ARCore's Geospatial API. Full detail is in the MCP Server Tool Reference, or ask your agent to call reactviro_get_platform_matrix.

Why Build with ReactVision

  • Agent first tools. The MCP Server and the Studio Agent, so your agent can build XR rather than guess at it.
  • Multi-user out of the box. A shared coordinate frame, replicated room state and smoothing, without writing a netcode layer.
  • Easy to adopt. It is React Native. If you know React, the learning curve is small.
  • Open source. MIT licensed renderer, no black box.
  • Infrastructure baked in. Cloud Anchors, Geospatial Anchors and co-location rooms without building them yourself.
  • Studio. A proper visual CMS for people who would rather not start in code.
  • MCP. Knowledge, validation, asset generation and testing, including environment testing, in one server.

Get Started

Next: Prerequisites to set up your environment.

Community and Support

Need help getting started, or want an expert to build with you?

Discord Banner 2


What’s Next

Ready to jump in? Move on to check the prerequisites. Need help? Consider hiring a ViroReact expert to help you get started.