Skip to content

About

Meta Wearables Device Access Toolkit for iOS

Resources

Code of conduct

Contributing

Security policy

Stars

591 stars

Watchers

23 watching

Forks

Repository files navigation

Meta Wearables Device Access Toolkit for iOS

Swift Package Docs

The Meta Wearables Device Access Toolkit enables developers to utilize Meta's AI glasses to build hands-free wearable experiences into their mobile applications. By integrating this SDK, developers can reliably connect to Meta's AI glasses and leverage capabilities like video and audio streaming and photo capture.

The Wearables Device Access Toolkit is in developer preview. Developers can access our SDK and documentation, test on supported AI glasses, and create organizations and release channels to share with test users.

Documentation & Community

Find our full developer documentation on the Wearables Developer Center.

You can find an overview of the Wearables Developer Center here. Create an account to stay informed of all updates, report bugs and register your organization. Set up a project and release channel to share your integration with test users.

For help, discussion about best practices or to suggest feature ideas visit our discussions forum.

See the changelog for the latest updates.

Including the SDK in your project

The Wearables Device Access Toolkit is distributed as a Swift Package.

  1. In Xcode, select File > Add Package Dependencies...
  2. Search for https://github.com/meta/meta-wearables-dat-ios in the top right corner
  3. Select meta-wearables-dat-ios
  4. Set the version to one of the available versions
  5. Click Add Package
  6. Select the target to which you want to add the packages
  7. Click Add Package

Developer Terms

  • By using the Wearables Device Access Toolkit, you agree to our Meta Wearables Developer Terms, including our Acceptable Use Policy.
  • By enabling Meta integrations, including through this SDK, Meta may collect information about how users' Meta devices communicate with your app. Meta will use this information collected in accordance with our Privacy Policy.
  • You may limit Meta's access to data from users' devices by following the instructions below.

Opting out of data collection

To configure analytics settings in your Meta Wearables DAT iOS app, you can modify your app's Info.plist file using either of these two methods:

Method 1: Using Xcode (Recommended)

  1. In Xcode, select your app target in the Project navigator
  2. Go to the Info tab
  3. Navigate to Custom iOS Target Properties and find the MWDAT key
  4. Add a new key under MWDAT called Analytics of type Dictionary
  5. Add a new key to the Analytics dictionary called OptOut of type Boolean and set the value to YES

Method 2: Direct XML editing

Add or modify the following in your Info.plist file.

<key>MWDAT</key>
<dict>
    <key>Analytics</key>
    <dict>
        <key>OptOut</key>
        <true/>
    </dict>
</dict>

Default behavior: If the OptOut key is missing or set to NO/<false/>, analytics are enabled (i.e., you are not opting out). Set to YES/<true/> to disable data collection.

Note: In other words, this setting controls whether or not you're opting out of analytics:

  • YES/<true/> = Opt out (analytics disabled)
  • NO/<false/> = Opt in (analytics enabled)

Crash reporting

The Wearables Device Access Toolkit can capture crashes originating from SDK code, store them locally, and chain with any existing crash handler your app installs.

Crash reporting is controlled by the analytics opt-out setting:

  • If analytics are enabled (no OptOut or OptOut = NO/<false/>) → Crash reporting is enabled
  • If analytics are disabled (OptOut = YES/<true/>) → Crash reporting is disabled

Use MWDAT > Analytics > OptOut to control both analytics and crash reporting.

AI-Assisted Development

This repository ships one public DAT knowledge base as a single plugin, plus file-based artifacts for the tools that do not use plugins:

Tool Public artifact Recommended setup
Muse Code, Claude Code, Codex plugins/mwdat-ios/ Add this GitHub repo as a marketplace, then install mwdat-ios
GitHub Copilot .github/copilot-instructions.md Auto-loaded by Copilot in VS Code
Cursor .cursor/rules/*.mdc Auto-loaded with glob-based triggers
AGENTS.md-compatible tools AGENTS.md Portable fallback for agents that read AGENTS.md
MCP-compatible editors https://mcp.developer.meta.com/wearables Connect as a remote HTTP MCP server; no authentication required
MockDevice MCP plugins/mwdat-ios/mcp/mockdevice/mockdevice_mcp_server.py Registered when you install the plugin in Muse Code, Claude Code, or Codex

Muse Code, Claude Code, and Codex install from the plugin payload under plugins/. Copilot, Cursor, and AGENTS.md readers use the native file-based artifacts at repo root.

The two MCP servers do different jobs. The remote server at mcp.developer.meta.com/wearables searches the published documentation. The MockDevice MCP runs locally and drives a simulated pair of glasses inside your app, so the assistant can verify its own changes without hardware. It needs Python 3.8+ and nothing else — no pip, uv, or node. Your app must start the MockDevice test server first by calling startTestServer; the CameraAccess sample does this in a debug build launched with --ui-testing. The MockDevice MCP sends nothing off your machine; the mock device events your app already logs record that the MCP drove them, and follow the same analytics opt-out. See the MockDevice MCP guide.

If you're an AI coding agent in a session that's already running, a plugin you install now may not load until the next session. Read AGENTS.md instead: it carries the same guidance as the plugin's skills.

Muse Code

muse plugins marketplace add mwdat-ios-marketplace https://github.com/meta/meta-wearables-dat-ios
muse plugins install mwdat-ios@mwdat-ios-marketplace
muse plugins approve mwdat-ios

Or use the helper script:

./install-skills.sh muse

Muse Code also reads AGENTS.md from your project root with no setup at all.

Claude Code

claude plugin marketplace add meta/meta-wearables-dat-ios
claude plugin install mwdat-ios@mwdat-ios-marketplace

Or use the helper script:

./install-skills.sh claude

Codex

codex plugin marketplace add meta/meta-wearables-dat-ios
codex plugin add mwdat-ios@mwdat-ios-marketplace

Or use the helper script:

./install-skills.sh codex

Other tool installs

Use the installer when you want the repo-native file surfaces for other tools:

./install-skills.sh copilot   # .github/copilot-instructions.md
./install-skills.sh cursor    # .cursor/rules/*.mdc
./install-skills.sh agents    # AGENTS.md
./install-skills.sh all       # Muse/Claude/Codex when available, plus Copilot/Cursor/AGENTS.md

Or run the helper remotely:

curl -sL https://raw.githubusercontent.com/meta/meta-wearables-dat-ios/main/install-skills.sh | bash

What's included

  • Getting started — SDK setup, SPM integration, Info.plist configuration
  • Camera streaming — Stream video frames and capture photos while video is active
  • Audio streaming — Receive synchronized PCM audio through Camera Stream (experimental; unavailable for production publishing)
  • Camera capture — Capture standalone high-quality photos with progress and typed errors (experimental; unavailable for production publishing)
  • Display — Use Display features on the Meta Ray-Ban Display glasses
  • Inputs — Receive navigation, button, capture, and drag interactions from glasses
  • Motion — Stream accelerometer, gyroscope, magnetometer, and orientation samples with explicit lifecycle control
  • Speech — Recognize on-device speech, handle partial/final results, and test with MockDeviceKit
  • Voice Invocations — Launch or activate an app from Hey Meta and acknowledge each action
  • MockDevice testing — Test without physical glasses using MockDeviceKit
  • MockDevice MCP — Drive a simulated device from the assistant to verify changes
  • Session lifecycle — Device session states, pause/resume, availability
  • Permissions & registration — App registration, camera permission flows
  • Debugging — Common issues, Developer Mode, version compatibility
  • Live debugging MCP — Read-only runtime diagnosis from app-visible DAT events
  • DAT docs MCP — Connect the hosted docs MCP server and search with search_dat_docs
  • Sample app guide — Building a complete DAT app

For static reference context, point your AI tool at the llms.txt index and let it follow the links to the page it needs. Installing the plugin registers https://mcp.developer.meta.com/wearables for live documentation search in Muse Code, Claude Code, and Codex. Other MCP-compatible editors can connect to that endpoint directly and use search_dat_docs. The public docs MCP server does not require authentication; do not configure tokens, OAuth, or custom authorization headers for it.

For hands-free verification, the bundled MockDevice MCP exposes mock_* tools that pair a device; set its power, wear, and hinge state; configure the camera feed; simulate capacitive touch, inputs, motion, and speech; set permission results; and read and tap the glasses display. Installing the plugin registers it in Muse Code, Claude Code, and Codex.

License

See the LICENSE file.

About

Meta Wearables Device Access Toolkit for iOS

Resources

Code of conduct

Contributing

Security policy

Stars

591 stars

Watchers

23 watching

Forks

Releases

Packages

Used by

Contributors

Languages