Google is expanding how AI agents can take part in Android development. The updates span Android Studio, command-line tools and APIs for apps, alongside new software and hardware for Android XR.
Android Studio opens to outside coding agents
With Android Studio Rabbit 2, developers can use the “Bring Your Own Agent” feature to run more than 40 coding agents through the open Agent Client Protocol. A built-in marketplace supports Anthropic’s Claude Agent, OpenAI’s Codex and Google Antigravity, which can be installed and used at the same time.
These agents can access development tools inside the IDE, including project structure, symbol resolution, the emulator and Logcat. The change gives developers options beyond Google’s own AI tools while placing agents alongside the resources they need to inspect and work on an Android project.
Google has also added Change Review to examine pending code changes. It can produce bug reports and explain modifications step by step, offering developers a way to inspect AI-generated edits. Android Studio Canary includes a migration assistant that translates older XML layouts into Jetpack Compose. Google is also testing Lightbuild, a declarative build system based on a YAML schema that people and agents can read and edit.
Google released version 2.0 of the Android Bench leaderboard to assess how models handle complex, multi-day development projects. Its measures include functionality, time spent and cost.
Terminal tools aim to speed up coding and debugging
The Android CLI is now an official command-line tool and an agent plugin for Claude Code, OpenAI’s Codex and Google Antigravity. Developers can use it from a terminal or in CI pipelines. Agents can render Compose previews in the terminal to check visual changes without starting an emulator.
In internal Google tests, agents completed UI tasks 77 percent faster and used up to three times fewer tokens with this preview feature. The CLI can also connect to streaming physical test devices from Google labs, helping agents investigate device-specific problems.
For memory analysis, Google integrated the open-source LeakCanary library into the CLI and Android Studio Profiler. Heap dumps are evaluated on the computer rather than the mobile device, keeping the app from slowing during testing. In a Google A/B test using Anthropic’s Claude Opus 5.5 model, the CLI tool helped find and fix memory leaks nearly twice as fast while cutting token usage by about 25 percent.
Other command-line tools let agents capture system traces with Android Performance Analyzer, retrieve crash reports from Firebase Crashlytics and check code optimization through the R8 compiler. Google is also providing over 30 Android Skills that supply documentation for tasks such as checking Google Play policies, addressing permission security issues and adapting layouts for foldables and laptops branded as Googlebooks.
Android apps can offer functions to system agents
Google is adding an alpha Agentic API to ML Kit for on-device agent features, powered by Gemini Nano 4. The interface can execute functions, save chat history, manage the orchestration loop and apply safety filters. This places parts of an agent’s work on the mobile device.
The beta AppFunctions API uses the Model Context Protocol as a common tool-calling layer, allowing apps to share functions with local or system-wide agents. An alpha Agent-to-UI interface for Jetpack Compose lets agents present interactive screens assembled from an app’s components instead of returning only text.
Google is also previewing the Connected Apps Platform, which lets developers connect app services to Google Search and the Gemini app through Android MCP or cloud-hosted MCP servers, managed from one console. Google says over 50 partner integrations, including companies like Adobe, Peloton and Experian, are already live in Gemini. They are set to roll out to web search in the coming weeks.
Android XR expands to glasses and larger screens
Updates to Jetpack Compose 1.13 and the AppState model are intended to help apps adapt to large screens. Google says the XR development environment is leaving the testing phase: the Jetpack XR SDK, Android Studio XR emulators and Engine Hub for Unity, Godot and Unreal are now stable releases.
Google and hardware maker XREAL unveiled the XREAL AURA, which Google XR chief Shahram Izadi says are the first wired smart glasses running Android XR. They weigh under 95 grams and have transparent, dimmable lenses with a 70-degree field of view. An external puck holds the computing unit and battery, doubles as a trackpad and runs on Qualcomm’s new Snapdragon Reality Elite chip.
Users can place up to five Android apps in view or connect a laptop by USB-C to mirror its screen. Gemini can see the surroundings and virtual screens, then organize windows, offer video game tips or highlight physical objects. The XREAL AURA is available to order in the US, Canada and South Korea for $1,099, with shipping set to begin in mid-November.