Google Expands Android Developer Tools with CLI Device Streaming and Wear OS M3 AI Skills

The landscape of Android development is evolving rapidly, with developers increasingly turning to artificial intelligence agents, large language models (LLMs), tools, and command-line interfaces (CLIs) to accelerate their workflows. Recognizing this shift, Google has announced major updates to its developer tooling ecosystem designed to streamline how high-quality Android applications are built, regardless of the specific agent or development environment preferred by the engineer.

The latest announcements center around significant enhancements to the Android CLI command-line tooling, most notably the integration of remote real-device access via Android Device Streaming. Alongside these terminal-centric capabilities, Google is expanding its repository of Android skills—structured instructions that ground AI models in official developer documentation—highlighting a specialized deep dive into the Wear OS Compose Material 3 skill. Together, these updates aim to bridge the gap between default LLM training cutoffs and platform-specific standards, offering a more robust, precise, and hardware-verified approach to AI-assisted mobile engineering.

Android CLI for Command-Line Development

At the core of the recent tooling updates is the Android CLI, a utility explicitly crafted to make command-line interface development smoother and more efficient for Android developers. Designed with flexibility in mind, the tool is built to support virtually any AI agent or external utility in constructing applications with greater efficiency.

When paired with the dedicated android-cli skill, AI agents gain the capability to leverage the Android CLI for a comprehensive suite of project management tasks. This includes creating, building, testing, and managing complex Android projects from the terminal. Furthermore, the tooling assists engineers and their automated assistants in setting up the local development environment, spinning up and executing emulators, and running extensive test suites without requiring constant manual intervention in a heavy graphical integrated development environment.

Android Device Streaming Now Available in CLI

One of the persistent challenges in mobile app development is ensuring that software performs reliably across a fragmented hardware landscape. While emulators are invaluable for rapid prototyping, real-world testing on physical devices remains critical for catching subtle, hardware- or operating system-specific bugs. However, developers do not always have physical access to the exact device variants required for thorough verification.

To address this limitation, Google has brought Android Device Streaming directly into the Android CLI, allowing developers and their AI agents to access real physical devices remotely. This integration means that an automated agent can now utilize Android Device Streaming from anywhere through the command-line interface, opening up new possibilities for automated testing and validation.

Through a secure ADB over SSL connection, an AI agent can interact with physical hardware just as if the device were plugged directly into the host machine via a USB cable. This terminal-driven pipeline allows the agent to spin up remote devices, deploy fresh builds, harvest logs and runtime traces, and even capture headless screenshots—all entirely within the terminal window.

To initiate the workflow, developers simply link their project, instruct their agent to query the list of available remote hardware, and select the target device required for the next phase of testing. This direct command-line access aims to drastically reduce the friction of hardware-in-the-loop testing for both human developers and autonomous coding assistants.

Grounding Agents with Android Skills

While large language models possess vast reservoirs of general programming knowledge, they frequently struggle with platform-specific standards, rapidly evolving APIs, and recent framework updates due to their inherent training data cutoffs. To bridge this gap between general model intelligence and official platform practices, Google continues to expand its Android skills repository.

Android skills are structured instructions encapsulated within standardized markdown files that actively ground AI agents using official guidance sourced directly from developer.android.com. Rather than relying on outdated model weights, these skills inject precise, fresh data, up-to-date API references, relevant code samples, and verified architectural patterns directly into the agent’s working context.

With more than twenty distinct skills currently available in the repository, developers can equip their AI agents to tackle increasingly complex and specialized mobile development tasks. The overarching philosophy and rigorous evaluation methodology behind these assets are detailed in Google’s documentation regarding how these resources are built to handle platform deprecations and standard updates.

Managing these skills across disparate projects and individual agent directories has been streamlined through the Android CLI. Because these skills are engineered to be entirely environment-agnostic, they function seamlessly across a wide variety of setups. Whether writing code natively within Android Studio, utilizing specialized environments like Antigravity, or pairing with third-party coding agents such as Claude and Codex, the Android skills are designed to integrate smoothly into any existing developer workflow.

Skill Spotlight: Wear Compose Material 3

Building applications for wearable devices running Wear OS presents unique design and technical challenges that differ significantly from standard smartphone development. Engineers must account for round viewports, rotary input mechanisms, specialized ambient display modes, stringent power consumption constraints, and specific UI paradigms such as the TransformingLazyColumn alongside AppScaffold and ScreenScaffolds containers.

Without explicit, context-aware guidance, standard LLMs frequently lack the nuanced understanding required to implement these distinct platform patterns correctly, often missing the details that make wearable applications polished and performant. To resolve this, Google has released the specialized Wear Compose Material 3 skill, targeting the wear/wear-compose-m3 namespace.

Early adopters across the industry have already reported substantial productivity gains after integrating this skill into their development pipelines. For instance, the engineering team at FotMob utilized the Wear Compose Material 3 skill to modernize their existing Wear M3-based application. Their migration tasks included updating multiple legacy lists to utilize TransformingLazyColumn with proper ScreenScaffold content padding, implementing ListHeader titles, applying SurfaceTransformation to cards and buttons, refining theme typography, and configuring accurate Wear previews.

The resulting code compiled successfully on the first pass and was rigorously verified on emulators for smooth scrolling, rotary input behavior, edge morphing, and right-to-left language support. This allowed the team to safely remove their legacy wrappers alongside a substantial accumulation of custom rotary and focus boilerplate code.

Notably, the skill successfully caught subtle implementation mistakes that the underlying AI model had initially overlooked, such as forgetting to forward the ScreenScaffold content padding into the internal list and defaulting to hardcoded sp values rather than proper theme typography. Reflecting on the efficiency of the workflow, Roy Solberg, Android Tech Lead at FotMob, noted that a single skill utilized over the course of a single afternoon successfully facilitated the migration of eight distinct lists while eliminating a pile of custom rotary code.