Meta Wearables Device Access Toolkit for Flutter

A Flutter plugin that brings Meta's Wearables Device Access Toolkit (DAT) to iOS and Android. Connect to Ray-Ban Meta, Oakley Meta, and Ray-Ban Display glasses — registration, live video streaming, photo capture, declarative on-glasses Display Access UI, the Mock Device Kit, and background streaming — all behind a single Dart API.

Wraps Meta's official DAT SDKs (v0.7.0) as binary dependencies. The DAT is in developer preview; apps cannot yet ship publicly via the App Store or Play Store. Create an organisation and release channel in the Wearables Developer Center to share builds with test users.

Unofficial. Not affiliated with, endorsed by, or officially connected to Meta Platforms, Inc. "Meta", "Ray-Ban Meta", "Oakley Meta", and "Ray-Ban Display" are trademarks of their respective owners.

Documentation & Community

Find Meta's full developer documentation on the Wearables Developer Center.

Plugin-specific guides live in doc/:

doc/getting_started.md — pubspec, iOS Info.plist, Android AndroidManifest.xml, Developer Mode.
doc/registration_flow.md — registration deep-link wiring.
doc/streaming.md — texture rendering, video codecs, photo capture, background streaming.
doc/frame_processing.md — opt-in per-frame videoFramesStream, recording, OCR/ML pipelines.
doc/display_access.md — declarative UI on Ray-Ban Display glasses (FlexBox/Text/Image/Button/Icon/VideoPlayer).
doc/mock_device.md — Mock Device Kit.
doc/troubleshooting.md — common pitfalls.

For help or to suggest feature ideas, open an issue on GitHub.

See the changelog for the latest updates.

Compatible devices

Ray-Ban Meta (Gen 1 and Gen 2)
Oakley Meta (HSTN, Vanguard)
Ray-Ban Display

A paired phone running the Meta AI companion app with Developer Mode enabled is required during the developer preview.

Including the SDK in your project

dependencies:
  meta_wearables_dat_flutter: ^0.7.1

flutter pub get

Enable Developer Mode in the Meta AI app (one-time, per phone)

STOP — do this before running anything. Until your app is approved in the Wearables Developer Center, registration is gated by a toggle inside the Meta AI mobile app. The manifest values below (MetaAppID = "0", CLIENT_TOKEN = "0") are only the code-side half of the handshake; without the phone-side toggle the SDK will fail with:

iOS: Internal error toast in Meta AI immediately after you tap Allow, your app never receives the deep link.

Android: RegistrationError(code: HTTP_REQUEST_FAILED, ...) with HTTP 401 from api2.ar.meta.com because the SDK falls back to real attestation.

Both symptoms disappear the instant the toggle is on.

Open the Meta AI app on the same phone you'll run your Flutter app on (the one paired with your glasses).
Tap your profile/avatar → Settings → scroll to the bottom → toggle Developer Mode on. (Older builds: Settings → Advanced → Developer Mode.)
Restart your Flutter app and try startRegistration() again.

The matching code-side switches are MetaAppID = "0" in your iOS Info.plist and APPLICATION_ID = "0" + CLIENT_TOKEN = "0" in your Android <meta-data> — already set in the snippets below.

iOS

Pick any short alphanumeric URL scheme for your app (no underscores — they break Meta's redirect-URL builder). Examples: mywearablesapp, metasdkdemo. The snippets below assume mywearablesapp; replace it everywhere it appears.

1. Enable Swift Package Manager (once per machine)

flutter config --enable-swift-package-manager

2. Raise the iOS deployment target to 17.0

The plugin's iOS SDK requires iOS 17 minimum. Set it in two places — they must match or the build fails:

ios/Podfile (uncomment / set the top line):

platform :ios, '17.0'

ios/Runner.xcodeproj → open ios/Runner.xcworkspace in Xcode → Runner target → General → Minimum Deployments → set iOS to 17.0.

Then refresh CocoaPods:

cd ios && pod install && cd ..

3. Add the `MWDAT` dict and related keys to `ios/Runner/Info.plist`

The DAT SDK validates the MWDAT dict as all-or-nothing attestation — all four keys (AppLinkURLScheme, MetaAppID, ClientToken, TeamID) must be present and non-empty or startRegistration() throws RegistrationError.configurationInvalid before Meta AI even opens. Paste this block inside the root <dict> of Info.plist:

<!-- Begin required section for Meta Wearables Device Access Toolkit. -->
<key>MWDAT</key>
<dict>
  <!-- MUST end with "://". Meta AI builds the callback URL by literally
       concatenating this value with "?authorityKey=...&metaWearablesAction=
       register&..."; without the separator the URL is malformed and iOS
       silently drops it. The scheme itself (without "://") must also be
       listed under CFBundleURLTypes below. RFC 3986 — alphanumeric, no
       underscores. -->
  <key>AppLinkURLScheme</key>
  <string>mywearablesapp://</string>

  <!-- "0" = Developer Mode sentinel. Pairs with the Meta AI app's
       Developer Mode toggle (see above). The SDK skips Wearables
       Developer Center attestation when MetaAppID is "0". -->
  <key>MetaAppID</key>
  <string>0</string>

  <!-- Required to be present and non-empty. The value is not validated
       in Developer Mode. For a published app, replace with the value
       from https://developers.meta.com/wearables/. -->
  <key>ClientToken</key>
  <string>developer-mode-placeholder</string>

  <!-- $(DEVELOPMENT_TEAM) is expanded by Xcode at build time from
       Signing & Capabilities. -->
  <key>TeamID</key>
  <string>$(DEVELOPMENT_TEAM)</string>

  <!-- Opt out of the SDK's analytics uploads to api2.ar.meta.com so a
       partial telemetry config in Developer Mode cannot fail
       registration with a misleading "Internal error". -->
  <key>Analytics</key>
  <dict>
    <key>OptOut</key>
    <true/>
  </dict>
</dict>

<!-- The OS routes the Meta AI deep-link callback to whichever app
     declares the matching scheme here. The string MUST match
     AppLinkURLScheme above, WITHOUT the trailing "://". -->
<key>CFBundleURLTypes</key>
<array>
  <dict>
    <key>CFBundleURLSchemes</key>
    <array>
      <string>mywearablesapp</string>
    </array>
  </dict>
</array>

<!-- The DAT SDK preflights `canOpenURL("fb-viewapp://...")` before
     it'll try to open Meta AI. Without `fb-viewapp` here iOS returns
     false and the SDK throws configurationInvalid (raw 1) before
     Meta AI even opens. -->
<key>LSApplicationQueriesSchemes</key>
<array>
  <string>fb-viewapp</string>
</array>

<!-- Streaming session requires `audio` to be declared even if you
     never start a stream during registration. The other three keep
     the BT + glasses transport alive in the background. -->
<key>UIBackgroundModes</key>
<array>
  <string>audio</string>
  <string>bluetooth-central</string>
  <string>bluetooth-peripheral</string>
  <string>external-accessory</string>
</array>

<key>NSBluetoothAlwaysUsageDescription</key>
<string>Needed to connect to Meta AI Glasses.</string>
<key>NSLocalNetworkUsageDescription</key>
<string>Allows your phone to find and connect to your glasses over Wi-Fi.</string>

<key>NSBonjourServices</key>
<array>
  <string>_bonjour._tcp</string>
</array>

<key>UISupportedExternalAccessoryProtocols</key>
<array>
  <string>com.meta.ar.wearable</string>
</array>
<!-- End required section for Meta Wearables Device Access Toolkit. -->

4. Forward Meta AI's deep-link callback (scene-based apps)

Flutter ≥ 3.32 generates a scene-based iOS lifecycle (a UIApplicationSceneManifest block in Info.plist and a ios/Runner/SceneDelegate.swift). On those apps iOS delivers the Meta AI callback URL to your host app's SceneDelegate, not to the plugin — and FlutterSceneDelegate does not auto-forward URLs to plugins. Replace ios/Runner/SceneDelegate.swift with:

import Flutter
import UIKit

class SceneDelegate: FlutterSceneDelegate {

  override func scene(
    _ scene: UIScene,
    willConnectTo session: UISceneSession,
    options connectionOptions: UIScene.ConnectionOptions
  ) {
    super.scene(scene, willConnectTo: session, options: connectionOptions)
    forward(urlContexts: connectionOptions.urlContexts)
  }

  override func scene(
    _ scene: UIScene,
    openURLContexts URLContexts: Set<UIOpenURLContext>
  ) {
    super.scene(scene, openURLContexts: URLContexts)
    forward(urlContexts: URLContexts)
  }

  private func forward(urlContexts: Set<UIOpenURLContext>) {
    for context in urlContexts {
      NotificationCenter.default.post(
        name: Notification.Name("MetaWearablesDatHandleURL"),
        object: nil,
        userInfo: ["url": context.url],
      )
    }
  }
}

The plugin subscribes to MetaWearablesDatHandleURL at registration time and routes the URL to the SDK natively — you do not call handleUrl(...) from Dart. If your app uses the classic AppDelegate lifecycle (no scene manifest), skip this step; the plugin auto-consumes the URL via application(_:open:options:).

Android

Pick the same URL scheme you used on iOS (without the :// suffix). The snippets below use mywearablesapp; replace it everywhere.

1. Make `MainActivity` extend `FlutterFragmentActivity`

android/app/src/main/kotlin/.../MainActivity.kt:

import io.flutter.embedding.android.FlutterFragmentActivity

class MainActivity : FlutterFragmentActivity()

Meta's camera-permission contract requires a ComponentActivity; FlutterFragmentActivity qualifies, FlutterActivity doesn't.

2. Set `minSdk = 31` and use NDK 28.2.13676358

In android/app/build.gradle.kts:

android {
    // Meta's `mwdat-core` AAR was built with NDK r27, but most Flutter
    // plugin transitive deps (e.g. `jni`) require r28.2. NDK is
    // backward compatible per AGP's "use highest" rule.
    ndkVersion = "28.2.13676358"

    defaultConfig {
        minSdk = 31
    }
}

3. Add Meta's GitHub Packages Maven repo

The Meta Android DAT SDK is published to GitHub Packages and requires a PAT with the read:packages scope.

In android/settings.gradle.kts, inside dependencyResolutionManagement { repositories { ... } }:

maven {
    url = uri("https://maven.pkg.github.com/facebook/meta-wearables-dat-android")
    credentials {
        username = "" // not needed; the PAT carries the user
        password = System.getenv("GITHUB_TOKEN")
            ?: providers.gradleProperty("github_token").orNull
            ?: ""
    }
}

Then either export GITHUB_TOKEN in your shell or add to android/local.properties:

github_token=ghp_yourPersonalAccessTokenWithReadPackagesScope

Create the PAT at github.com/settings/tokens with scope read:packages only.

4. Add `MWDAT` meta-data, permissions, and the deep-link intent-filter

android/app/src/main/AndroidManifest.xml:

<manifest xmlns:android="http://schemas.android.com/apk/res/android">

  <!-- Required by the Meta Wearables DAT SDK. -->
  <uses-permission android:name="android.permission.BLUETOOTH" />
  <uses-permission android:name="android.permission.BLUETOOTH_CONNECT" />
  <uses-permission android:name="android.permission.INTERNET" />

  <application
      android:name="${applicationName}"
      android:label="Your App"
      android:icon="@mipmap/ic_launcher">

    <!-- Both values must be "0" in Developer Mode (per Meta's official
         Android docs). An empty string causes TOKEN_NOT_CONFIGURED;
         any non-"0" placeholder makes the SDK attempt a real HTTP
         attestation against api2.ar.meta.com and registration fails
         with HTTP 401. For production, replace with the AppID and
         ClientToken from https://developers.meta.com/wearables/.
         DAM_ENABLED tells the SDK you are running in Developer Access
         Mode — without it `usesDam` stays false and the session manager
         rejects every paired device with "No eligible device found".
         ANALYTICS_OPT_OUT prevents the SDK from uploading developer
         analytics during testing. -->
    <meta-data
        android:name="com.meta.wearable.mwdat.APPLICATION_ID"
        android:value="0" />
    <meta-data
        android:name="com.meta.wearable.mwdat.CLIENT_TOKEN"
        android:value="0" />
    <meta-data
        android:name="com.meta.wearable.mwdat.DAM_ENABLED"
        android:value="true" />
    <meta-data
        android:name="com.meta.wearable.mwdat.ANALYTICS_OPT_OUT"
        android:value="true" />

    <activity
        android:name=".MainActivity"
        android:exported="true"
        android:launchMode="singleTop"
        android:theme="@style/LaunchTheme"
        android:configChanges="orientation|keyboardHidden|keyboard|screenSize|smallestScreenSize|locale|layoutDirection|fontScale|screenLayout|density|uiMode"
        android:hardwareAccelerated="true"
        android:windowSoftInputMode="adjustResize">

      <intent-filter>
        <action android:name="android.intent.action.MAIN"/>
        <category android:name="android.intent.category.LAUNCHER"/>
      </intent-filter>

      <!-- Registration callback from the Meta AI app. The scheme MUST
           match AppLinkURLScheme on iOS and the scheme you pick for
           your app. `launchMode="singleTop"` above is required so the
           SDK can find and route the inbound intent. -->
      <intent-filter>
        <action android:name="android.intent.action.VIEW" />
        <category android:name="android.intent.category.BROWSABLE" />
        <category android:name="android.intent.category.DEFAULT" />
        <data android:scheme="mywearablesapp" />
      </intent-filter>
    </activity>
  </application>
</manifest>

The plugin's manifest already merges in FOREGROUND_SERVICE, FOREGROUND_SERVICE_CONNECTED_DEVICE, WAKE_LOCK, and POST_NOTIFICATIONS plus the background-streaming <service> entry, so you don't have to declare them.

Integration lifecycle

import 'package:meta_wearables_dat_flutter/meta_wearables_dat_flutter.dart';

// 1. Permissions (Bluetooth/Internet on Android, no-op on iOS).
await MetaWearablesDat.requestAndroidPermissions();

// 2. Register with the Meta AI app via deep link. `appId` and
//    `urlScheme` are read from the host app's Info.plist / Android
//    meta-data — no need to repeat them in Dart.
await MetaWearablesDat.startRegistration();

// 3. Camera permission (Meta AI bottom sheet).
await MetaWearablesDat.requestCameraPermission();

// 4. Start streaming and render the texture.
final textureId = await MetaWearablesDat.startStreamSession();
// Texture(textureId: textureId)

// 5. Capture stills, observe state.
final photo = await MetaWearablesDat.capturePhoto();
MetaWearablesDat.streamSessionStateStream().listen(print);

See samples/camera_access/ for a complete integration that mirrors Meta's official iOS and Android CameraAccess samples, and samples/display_access/ for the Display Access "Car Maintenance" tutorial on Ray-Ban Display glasses.

Developer Terms

By using the Wearables Device Access Toolkit, you agree to Meta's Meta Wearables Developer Terms, including the Acceptable Use Policy.
By enabling Meta integrations through this plugin, Meta may collect information about how users' Meta devices communicate with your app. Meta uses this information in accordance with the Meta Privacy Policy.
You may limit Meta's access to data from users' devices by opting out of analytics as described below.

Opting out of data collection

This plugin is a thin bridge — analytics are controlled by Meta's underlying SDKs and you opt out exactly as you would in a native project.

iOS — add an Analytics.OptOut key inside the MWDAT dict in ios/Runner/Info.plist:

<key>MWDAT</key>
<dict>
  <key>Analytics</key>
  <dict>
    <key>OptOut</key>
    <true/>
  </dict>
  <!-- other MWDAT keys ... -->
</dict>

Android — add the matching meta-data entry inside android/app/src/main/AndroidManifest.xml:

<meta-data
  android:name="com.meta.wearable.mwdat.ANALYTICS_OPT_OUT"
  android:value="true" />

Default behavior: if the key is missing or false, analytics are enabled. Set it to true to disable data collection.

AI-Assisted Development

This repository ships config for three AI coding assistants, all generated from the same canonical knowledge in AGENTS.md:

Tool	Config	How it loads
Claude Code	`.claude/skills/*.md`	Auto-discovered when you open the project
GitHub Copilot	`.github/copilot-instructions.md`	Auto-loaded by Copilot in VS Code
Cursor	`.cursor/rules/*.mdc`	Auto-loaded with glob-based triggers

Quick setup

Install config for your preferred tool:

./install-skills.sh claude    # Claude Code only
./install-skills.sh copilot   # GitHub Copilot only
./install-skills.sh cursor    # Cursor only
./install-skills.sh agents    # AGENTS.md only
./install-skills.sh all       # All tools

Or install everything remotely with a single command:

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

If you cloned this repository, the config is already included — no setup needed.

What's included

Getting started — pubspec wiring, Info.plist, AndroidManifest.xml, Developer Mode.
Camera streaming — texture path, video codecs, photo capture, videoFramesStream.
Display access — declarative on-glasses UI, callbacks, video.
Mock device testing — MockDeviceKit from Dart.
Session lifecycle — DeviceSession vs StreamSession, pause/resume.
Permissions & registration — deep-link callbacks, camera permission flow.
Debugging — registration errors, no eligible device, Maven 401, SPM not enabled.
Sample app guide — building a complete Flutter DAT app.

For Meta's full API reference, point your AI tool at the llms.txt endpoint.

License

Acknowledgments

Built on top of Meta's official open-source SDKs:

Architecture inspiration (texture bridge, per-frame stream) drawn from the community flutter_meta_wearables_dat plugin. See NOTICE for full attribution.

Meta Wearables Device Access Toolkit for Flutter

Documentation & Community

Compatible devices

Including the SDK in your project

Enable Developer Mode in the Meta AI app (one-time, per phone)

iOS

1. Enable Swift Package Manager (once per machine)

2. Raise the iOS deployment target to 17.0

4. Forward Meta AI's deep-link callback (scene-based apps)

Android

1. Make `MainActivity` extend `FlutterFragmentActivity`

2. Set `minSdk = 31` and use NDK 28.2.13676358

3. Add Meta's GitHub Packages Maven repo

4. Add `MWDAT` meta-data, permissions, and the deep-link intent-filter

Integration lifecycle

Developer Terms

Opting out of data collection

AI-Assisted Development

Quick setup

What's included

License

Acknowledgments

Libraries

meta_wearables_dat_flutter package

Meta Wearables Device Access Toolkit for Flutter

Documentation & Community

Compatible devices

Including the SDK in your project

Enable Developer Mode in the Meta AI app (one-time, per phone)

iOS

1. Enable Swift Package Manager (once per machine)

2. Raise the iOS deployment target to 17.0

3. Add the MWDAT dict and related keys to ios/Runner/Info.plist

4. Forward Meta AI's deep-link callback (scene-based apps)

Android

1. Make MainActivity extend FlutterFragmentActivity

2. Set minSdk = 31 and use NDK 28.2.13676358

3. Add Meta's GitHub Packages Maven repo

4. Add MWDAT meta-data, permissions, and the deep-link intent-filter

Integration lifecycle

Developer Terms

Opting out of data collection

AI-Assisted Development

Quick setup

What's included

License

Acknowledgments

Libraries

meta_wearables_dat_flutter package

3. Add the `MWDAT` dict and related keys to `ios/Runner/Info.plist`

1. Make `MainActivity` extend `FlutterFragmentActivity`

2. Set `minSdk = 31` and use NDK 28.2.13676358

4. Add `MWDAT` meta-data, permissions, and the deep-link intent-filter