AI-Assisted Development

Updated: Mar 12, 2026

Overview

The Wearables Device Access Toolkit provides two levels of AI assistance for developers:
  1. Git-native tool integrations — SDK knowledge delivered directly from the DAT GitHub repos through Claude marketplace/plugin installs, Codex plugins, and repo-native tool files.
  2. API reference endpoint — The full API surface served via llms.txt⁠ as a supplementary reference for on-demand queries.

Git-native tool integrations

The SDK GitHub repos ship the same public DAT guidance in several formats. Each tool gets the same core knowledge — setup guides, streaming patterns, MockDeviceKit testing, session lifecycle, permissions, debugging, and sample app guidance — in whatever format it expects.
ToolPublic artifactRecommended setup
Claude Code⁠
.claude-plugin/marketplace.json + plugin manifests
Add the DAT GitHub repo as a marketplace, then install the DAT plugin
Codex
.codex-plugin/plugin.json inside plugins/
Install the DAT plugin from a cloned DAT GitHub repo
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
AGENTS.md
Portable fallback for tools that read AGENTS.md
See the dedicated setup guides for each tool: Claude Code, Codex, GitHub Copilot, Cursor, AGENTS.md.

Setup

For Claude Code and Codex, start from the DAT GitHub repo for your platform:
claude plugin marketplace add facebook/meta-wearables-dat-ios
claude plugin install mwdat-ios@mwdat-ios-marketplace

git clone https://github.com/facebook/meta-wearables-dat-ios.git
cd meta-wearables-dat-ios
codex plugin install ./plugins/mwdat-ios
Android uses the same flow with facebook/meta-wearables-dat-android and mwdat-android.
For tools that use repo-native checked-in files instead of plugin manifests, use:
./install-skills.sh copilot
./install-skills.sh cursor
./install-skills.sh agents

What the config covers

All surfaces receive the same SDK knowledge, covering:
  • Getting started — Dependencies, project setup, and first integration
  • Session lifecycle — Connecting, reconnecting, and managing device sessions
  • Camera streaming — Resolution selection, frame rates, and photo capture
  • Permissions — Requesting and handling device permissions
  • MockDeviceKit — Testing without physical hardware
  • Debugging — Common issues, Developer Mode, version compatibility
  • Sample app guide — Building a complete DAT app from scratch

API reference endpoint (llms.txt)

As a supplement to the GitHub-hosted integrations, the SDK provides an llms.txt⁠ endpoint with the full API surface. The GitHub integrations cover patterns and best practices; the endpoint covers the API reference.

Endpoints

EndpointDescription
wearables.developer.meta.com/llms.txt⁠
Index of available documentation sections with links
wearables.developer.meta.com/llms.txt?full=true⁠
Full API reference content in a single document
The index endpoint follows the llms.txt specification⁠ and returns a lightweight listing of documentation sections. The ?full=true variant returns the complete API reference inline, which is what most AI tools need to write code.

What’s included

The endpoint serves API reference documentation for both iOS (Swift) and Android (Kotlin) platforms, covering:
  • MWDATCore — App registration, device discovery, session management, and telemetry
  • MWDATCamera — Camera access, resolution and frame rate selection, and photo capture
  • MWDATMockDevice — Simulated device for testing without physical hardware

Tips for effective use

  • Start with the GitHub-hosted integration — Use the Claude marketplace or Codex plugin when your tool supports it. Use the repo-native file installs for Copilot, Cursor, or AGENTS.md readers.
  • Add the API reference when you need specifics — If your AI tool can’t find a particular method signature or parameter type, point it at the llms.txt endpoint for the full API surface.
  • Be specific in your prompts — Mention the platform (iOS or Android) and the module you’re working with (MWDATCore, MWDATCamera, or MWDATMockDevice).
  • Combine with the guides — For deeper integration patterns and lifecycle management, point your AI tool at the integration overview and platform-specific integration guides (iOS, Android).