Google is letting developers run third-party AI coding agents inside Android Studio, its Android app development environment. The new “Bring Your Own Agent” feature in Android Studio Rabbit 2 uses the open Agent Client Protocol and supports more than 40 agents, including Anthropic’s Claude Agent, OpenAI’s Codex and Google Antigravity.
The agents can access Android Studio tools such as a project’s structure, code symbols, the emulator and Logcat, which displays app logs. Google also added Change Review, a tool that examines proposed code changes, reports possible bugs and explains edits. A migration assistant that converts older XML layouts to Jetpack Compose is in the Android Studio Canary preview; Lightbuild, a proposed build system, is being tested.
Tools for agents in the terminal
Google’s Android command-line tool can now be used as an agent plugin for Claude Code, Codex and Antigravity. Agents can render Compose screen previews in a terminal to check interface changes without starting an emulator. In internal tests, Google says agents completed UI tasks 77% faster and used up to three times fewer tokens with this feature.
The CLI and Android Studio Profiler also integrate LeakCanary, an open-source tool for finding memory leaks. Google says its A/B test with Anthropic’s Claude Opus 5.5 found and fixed leaks nearly twice as fast using the CLI, while cutting token use by about 25%. Google has also released more than 30 Android Skills, which provide agents with guidance on tasks such as app permissions and Google Play policies.
New app APIs and smart glasses
For developers building AI features into apps, Google has introduced an alpha Agentic API for ML Kit, powered by Gemini Nano 4. It handles function calls, chat history and safety checks on the device. AppFunctions, a beta API, lets apps share functions with agents using the Model Context Protocol, while an alpha Agent-to-UI interface lets agents create interactive screens from app components.
Google also announced the XREAL AURA, wired smart glasses running Android XR. They can show up to five app windows and connect to a laptop by USB-C. Google says Gemini can interpret both the wearer’s surroundings and virtual screens. The glasses weigh under 95 grams and are available to order in the US, Canada and South Korea for $1,099, with shipping due to begin in mid-November.
Why it matters
Android developers can use coding agents from Anthropic and OpenAI in Android Studio rather than being limited to Google’s own tools. Google’s performance figures come from its own tests, so developers should not treat them as independent proof that these tools will speed up every project.
