Google opens Android Studio to more than 40 coding agents, including Claude and Codex

Android developers can now bring Claude, Codex and dozens of other coding agents into Google’s app-building software, with access to tools for running and checking their work. Google released Android Studio Rabbit 2 as a stable download, pairing its Bring Your Own Agent feature with AI-assisted code review and expanded terminal integrations.

An open connection to the development tools

Announced at Android Dev Summit on October 28, the release supports more than 40 coding agents through the Agent Client Protocol, or ACP. This open standard connects AI agents to code editors. It gives Android Studio a common connection method rather than requiring a separate custom integration for each compatible agent.

Developers can find the agent marketplace under Settings > Tools > AI > Agents. Available choices include Anthropic’s Claude Agent, OpenAI’s Codex and Google Antigravity. They can install multiple agents and run them simultaneously, rather than choosing a single assistant for the environment.

The integration reaches beyond a chat panel. Agents receive access to development tools that let them inspect a project and check how an app behaves:

• Project structure, builds and symbol resolution, which helps identify what names in the code refer to.
• The Android Emulator, a virtual device where an agent can run the app.
• Logcat, the running log of app activity that can help an agent check its work.

A review step for agent-written changes

Google is also addressing the work developers face after an agent edits a project. The company acknowledges that large agent-driven changes can be difficult to review and understand. Its new Change Review feature is an AI-assisted reviewer that scans pending changes, including those produced by coding agents.

The tool analyzes modifications, generates bug reports and explains changes step by step. For longer edits, it can walk developers through what changed, giving them a way to inspect an agent’s output rather than only receiving the finished code.

Terminal agents get visual checks and memory diagnostics

For developers working outside the editor, Google expanded the Android CLI, its command-line toolkit for building and testing apps. It is now available as an official agent plugin for Claude Code, OpenAI Codex and Google Antigravity, supporting agents used in terminals or automated development pipelines.

New commands render Compose interface previews directly in the terminal. Agents can visually check interface changes without starting an emulator. Google says internal tests found agents completed these interface tasks 77% faster and used up to three times fewer tokens—the units of text AI systems process—with the preview feature.

Memory debugging gets another integration. Google built the open-source LeakCanary library into both the CLI and Android Studio Profiler. It detects memory leaks, where an app holds on to memory it no longer needs. The system analyzes memory snapshots on the computer instead of the mobile device, avoiding a testing slowdown on the app itself.

That result has a substantial limit: each setup was run only once, according to Digital Trends. It is a company-run demonstration, not a repeated comparison establishing how much time developers should expect to save across other apps or debugging tasks.