Glonce

Google opens Android Studio to more than 40 outside AI agents, adds Gemini Nano 4 agent API

toolsThe Glonce editorsThe Decoder🔥72Glonce rating72Impact66Novelty62Relevance100Surprise40Players90Credibility82

At this year's Android Dev Summit, Google introduced updates to its developer tools and system APIs, with the opening of Android Studio as the main focus. Android Studio Rabbit 2 launches a "Bring Your Own Agent" (BYOA) feature. Developers are no longer limited to Google's own AI tools: through the open Agent Client Protocol (ACP) they can run more than 40 different coding agents directly in the IDE. Anthropic's Claude Agent, OpenAI's Codex and Google Antigravity can be installed and run at the same time through a built-in marketplace, and they get direct access to IDE tools such as project structure, symbol resolution, the emulator and Logcat.

Google also released version 2.0 of the Android Bench leaderboard, which evaluates language models on multi-day development projects by functionality, time spent and cost. Because AI-generated changes often get messy, Android Studio gains a verification tool called Change Review, which analyzes pending changes, generates bug reports and explains modifications step by step. The Canary preview adds a migration assistant that automatically translates older XML layouts into Jetpack Compose. Google is also testing Lightbuild, a declarative build system based on a YAML schema that both humans and AI agents can read and edit.

For developers who run agents in the terminal or in CI pipelines, Google expanded the Android CLI. It is now an official agent plugin for Claude Code, OpenAI's Codex and Google Antigravity. New commands let agents render Compose previews in the terminal and check UI changes without booting an emulator; in internal Google tests, agents finished UI tasks 77 percent faster and used up to three times fewer tokens. The CLI can also stream physical test devices from Google labs so agents can investigate device-specific bugs. Google built the open-source LeakCanary library into both the CLI and the Android Studio Profiler, evaluating heap dumps on the computer instead of the phone so the app does not slow down during testing. In a Google A/B test with Anthropic's Claude Opus 5.5, the tool helped find and fix memory leaks nearly twice as fast while cutting token usage by about 25 percent. Other CLI tools let agents capture system traces via the Android Performance Analyzer, pull crash reports from Firebase Crashlytics, or check code optimization through the R8 compiler. Since language models often lack training on the newest APIs, Google now provides over 30 Android Skills, documentation fed to agents for jobs like checking Google Play policies before submission, handling permission security issues, or adapting layouts for foldables and laptops branded as Googlebooks.

On the app side, Google is adding an alpha Agentic API powered by Gemini Nano 4 to ML Kit for local on-device processing. It automatically executes functions, saves chat history, manages the orchestration loop and provides safety filters. The AppFunctions API launches in beta, using the Model Context Protocol (MCP) as a unified tool-calling layer so apps can share functions with local or system-wide agents. An alpha Agent-to-UI (A2UI) interface for Jetpack Compose builds on it, letting agents show interactive interfaces built from an app's own components instead of plain text. To bring app services into Google Search and the Gemini app, Google is launching a preview of the Connected Apps Platform: developers link apps via Android MCP or cloud-hosted MCP servers and manage them in a single console. Google says over 50 partner integrations from companies like Adobe, Peloton and Experian are already live in the Gemini app and will roll out to web search in the coming weeks.

For other form factors, updates to Jetpack Compose 1.13 and the AppState model ease adapting apps to large screens. The Jetpack XR SDK, the XR emulators in Android Studio and the Engine Hub for Unity, Godot and Unreal are now stable. With hardware maker XREAL, Google unveiled the XREAL AURA, which 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. The computing unit and battery sit in an external puck that doubles as a trackpad and runs on Qualcomm's new Snapdragon Reality Elite chip. Users can place up to five Android apps in view at once or connect a USB-C cable to mirror a laptop screen, and Gemini can see both the real surroundings and the virtual screens to organize windows, offer game tips or highlight physical objects. The glasses can be ordered now in the US, Canada and South Korea for $1,099, with shipping set to begin in mid-November.

Key facts

  • Android Studio Rabbit 2 adds Bring Your Own Agent: via the open Agent Client Protocol, more than 40 coding agents, including Claude Agent, Codex and Google Antigravity, run in the IDE with access to the emulator, Logcat and symbol resolution.
  • In internal Google tests, rendering Compose previews in the terminal made agents finish UI tasks 77 percent faster with up to three times fewer tokens; a LeakCanary A/B test with Claude Opus 5.5 found and fixed memory leaks nearly twice as fast with about 25 percent fewer tokens.
  • New agent APIs: an alpha Agentic API on Gemini Nano 4 in ML Kit, AppFunctions in beta over MCP, A2UI in alpha for Jetpack Compose, and a Connected Apps Platform preview; over 50 partner integrations are live in the Gemini app.
  • Google provides over 30 Android Skills, and the Android CLI is now an official agent plugin for Claude Code, Codex and Google Antigravity.
  • Google and XREAL unveiled the XREAL AURA Android XR smart glasses at $1,099, orderable in the US, Canada and South Korea, shipping from mid-November.

Why it matters

Android's main IDE is no longer tied to Google's own AI tooling. By building on the open Agent Client Protocol, Google lets developers run agents from Anthropic and OpenAI next to its own Antigravity, and ships the Android CLI as an official plugin for Claude Code, Codex and Antigravity. The same announcement pushes agents into the apps themselves: Gemini Nano 4 on-device, MCP-based tool calling, and routes into Google Search and the Gemini app.

Who it affects

Android developers who already work with coding agents such as Claude Agent, Codex or Antigravity, and teams that run agents in the terminal or in CI pipelines. App makers who want their functions callable by local or system-wide agents are also in scope, as are XR developers (the Jetpack XR SDK and XR emulators are now stable) and anyone targeting large screens, foldables and Googlebooks. Partners such as Adobe, Peloton and Experian already have integrations live in the Gemini app.

How to use it

Install Android Studio Rabbit 2 and pick agents from the built-in marketplace; several can run at once. For terminal or CI work, use the Android CLI, which now offers Compose previews without an emulator, LeakCanary memory analysis, system traces, Crashlytics reports and R8 checks. Android Skills supply current documentation to agents. For app features, the Gemini Nano 4 Agentic API (alpha) sits in ML Kit, AppFunctions is in beta, A2UI is in alpha, and the Connected Apps Platform is a preview where apps link via Android MCP or cloud-hosted MCP servers and are managed in one console. The XREAL AURA costs $1,099 and ships from mid-November.

How solid is it

This is a first-party announcement, relayed by The Decoder. The speed and token figures (77 percent faster, up to three times fewer tokens, nearly twice as fast on memory leaks, about 25 percent fewer tokens) come from internal Google tests, and the source gives no methodology or independent verification. Several pieces are early: the Agentic API and A2UI are alpha, AppFunctions is beta, the Connected Apps Platform is a preview, the migration assistant is in Canary, and Lightbuild is only being tested. No Android Bench 2.0 results or rankings are given.

Risks and caveats

Much of the new tooling is pre-release, so interfaces may change. The source gives no pricing for the Android Studio agents, the CLI or the Agentic API, and does not say which Gemini Nano 4 devices will support the Agentic API. Google itself notes that AI-generated code changes often get messy, which is why it adds Change Review. The partner integrations reach web search only in the coming weeks, and XREAL AURA availability beyond the US, Canada and South Korea is not stated. No reaction from Anthropic or OpenAI is reported.