Google Updates Android CLI with Remote Device Streaming and Expands AI Agent Skills for Developers

As the landscape of mobile development continues to evolve alongside advancements in artificial intelligence, software engineers find themselves utilizing an increasingly diverse array of agents, large language models, tools, and command-line interfaces. Recognizing this flexibility, the Android development team has announced a series of significant updates designed to streamline application creation regardless of a developer’s preferred workflow. The latest announcements focus on expanding the capabilities of the Android command-line tooling, introducing remote hardware access directly into terminal-based workflows, and enriching the repository of structured instructions that help AI systems better understand modern Android and Wear OS standards.

The core objective behind these updates remains helping developers build high-quality applications efficiently. By refining the interaction between modern coding assistants and native Android infrastructure, the development team aims to bridge the gap between generalized artificial intelligence models and the specific, rapidly changing requirements of the Android ecosystem.

Android CLI for Command-Line Development

At the heart of terminal-centric and AI-assisted workflows is the Android CLI, a utility built to simplify command-line interface development. Designed to support virtually any AI agent or external developer tool, the CLI functions as a bridge that enables automated and manual systems to interact with Android projects more efficiently.

When paired with the dedicated android-cli skill, AI agents can leverage the command-line tool to handle a comprehensive range of project lifecycle tasks. These include creating, building, testing, and managing complex Android codebases, as well as assisting with development environment setup, emulator provisioning and execution, and automated test runs. By centralizing these operations within a unified command-line interface, developers can seamlessly integrate automated agents into their daily coding routines without sacrificing control over their local environment.

Android Device Streaming Now Available in CLI

One of the most persistent challenges in mobile engineering is ensuring that applications perform reliably across a fragmented ecosystem of physical hardware. While emulators provide a fast feedback loop during early development, catching hardware- and operating-system-specific issues ultimately requires testing on real physical devices. However, maintaining a comprehensive physical device lab in-house is often impractical or cost-prohibitive.

To address this limitation, Android Device Streaming has been integrated directly into the Android CLI, bringing remote access to physical devices straight to the terminal. This integration allows AI agents to utilize Android Device Streaming from anywhere within a command-line environment, opening up new possibilities for automated testing and remote debugging.

Through a secure Android Debug Bridge over Secure Sockets Layer connection, an AI agent can interact with physical devices as though they were physically plugged into the host machine via USB. This secure connection permits a wide range of headless operations, including spinning up remote devices, deploying fresh builds, collecting detailed system logs and execution traces, and capturing screenshots directly from the terminal interface.

Getting started with the feature involves linking a project, instructing an agent to list available remote devices, and selecting the desired hardware configuration for testing. This capability ensures that both developers and their automated assistants can verify software behavior on actual consumer hardware without maintaining physical devices on their desks.

Grounding Agents with Android Skills

While large language models possess vast amounts of generalized training data, they frequently struggle with platform-specific standards, newly introduced APIs, and rapid ecosystem updates. To mitigate this issue and prevent hallucinations or outdated code generation, the Android team has continued to expand its repository of Android skills.

Android skills consist of structured instructions, organized as dedicated files, that ground artificial intelligence agents using official guidance sourced directly from official developer documentation. Rather than relying solely on a model’s static training cutoff, these skills inject precise, up-to-date data, official API references, code samples, and modern architectural patterns directly into the agent’s active context.

With more than twenty skills currently available in the public repository, developers can equip their AI coding assistants to handle complex and specialized development domains with greater accuracy. The architecture behind these skills is thoroughly evaluated to ensure reliability. Furthermore, Android skills are entirely environment-agnostic, meaning they function smoothly across a wide variety of development setups—whether an engineer is writing code inside Android Studio, utilizing specialized integrated environments, or pairing with third-party coding agents. Managing these skills across different projects and agent directories is handled directly through the Android CLI, keeping the setup process straightforward and unobtrusive.

Skill Spotlight: Wear Compose Material 3

Developing applications for Wear OS presents unique design and engineering challenges that differ significantly from handheld mobile development. Engineers must account for round viewports, rotary input mechanisms, specialized ambient display modes, strict power consumption limits, and specific layout components such as transforming lazy columns and dedicated container structures.

Without explicit, context-aware guidance, generalized large language models frequently lack the nuanced understanding required to implement these distinct patterns correctly. To solve this problem, the Android team introduced the Wear Compose Material 3 skill, which provides agents with the precise knowledge needed to build modern watch applications.

Early adopters of the new skill have reported substantial productivity gains. For instance, the engineering team at FotMob utilized the Wear Compose Material 3 skill to modernize their existing application built on earlier Wear Material iterations. The migration tasks included updating multiple lists to use transforming lazy columns with proper content padding, incorporating list header titles, applying surface transformations to cards and buttons, adopting theme typography, and configuring reliable Wear previews.

The resulting code compiled successfully and was rigorously verified on emulators for smooth scrolling, rotary input response, edge morphing, and right-to-left language support. This automated and guided assistance allowed the development team to eliminate legacy wrappers and remove extensive custom rotary and focus boilerplate code. Crucially, the skill successfully caught subtle implementation mistakes that the underlying model would have otherwise missed, such as failing to forward content padding correctly into lists and mistakenly using hard-coded text dimensions instead of theme typography. As noted by the Android tech lead at FotMob, applying the skill streamlined an extensive migration process into a single productive afternoon, successfully removing piles of custom boilerplate code.