Google opens Android Studio to 40+ rival AI agents including Claude and Codex

Android Studio Rabbit 2 now lets you run rival AI agents directly inside your IDE, shifting your job from typing code to judging it.

Tools · Source: FoneArena

What happened

Google just released Android Studio Rabbit 2 at the Android Dev Summit 2026. The major update is a Bring Your Own Agent feature. Developers can now run over 40 external coding agents directly inside the IDE. Supported options include Anthropic's Claude, OpenAI's Codex, and Google Antigravity. These agents connect via the open Agent Client Protocol. They get direct access to core IDE tools like the project structure, symbol resolution, the emulator, and the Logcat system.

Google also updated the Android CLI, making it an official plugin for Claude Code and Codex. Agents can now render Jetpack Compose previews right in the terminal without booting an emulator. To manage the messy code these models often spit out, Google added a Change Review tool. It analyzes pending code changes, generates bug reports, and explains modifications step by step. Google also released version 2.0 of the Android Bench leaderboard to measure AI model performance on complex, multi-day development projects.

Beyond developer tools, Google launched new APIs to connect apps to system agents. The AppFunctions API is now in beta, acting as a unified tool-calling layer. An alpha Agentic API powered by Gemini Nano 4 brings local on-device processing to the ML Kit. Google also partnered with hardware maker XREAL to unveil the XREAL AURA, a pair of wired smart glasses running Android XR, marking the stable release of Android XR development tools.

Key facts

Why it matters

The developer workflow is fundamentally changing. You are no longer locked into Google's proprietary AI ecosystem to build Android apps. By opening the IDE to rival models, Google is turning Android Studio into a neutral battleground for the best AI tools. Your daily job shifts from writing boilerplate code to orchestrating and reviewing AI output. You are now an editor, not just a writer. The AI handles the syntax, while you handle the architecture and user experience.

The second-order effect is a massive reduction in development friction and API token costs. Internal Google tests showed agents finished UI tasks 77 percent faster and used up to three times fewer tokens when rendering previews in the terminal. Tools like LeakCanary integrated into the CLI also helped models like Claude Opus 5.5 find memory leaks twice as fast while cutting token usage by 25 percent. Building and debugging Android apps just got significantly cheaper and faster. This lowers the barrier to entry for solo founders and small engineering teams trying to ship complex mobile products.

For builders

Cut token costs with terminal previews

Agents can now render Compose previews directly in the terminal. This bypasses the need to boot an emulator. It saves time and drastically reduces API token usage for founders paying for AI models.

Audit AI code with Change Review

AI-generated code gets messy quickly. Use the new Change Review tool to automatically analyze pending changes and generate bug reports. It explains modifications step by step before you merge anything into production.

Expose app features to system agents

The new AppFunctions API beta lets you share your app functionality with local or system-wide agents. Early adopters can plug their services directly into Google Search and the Gemini app to reach more users.

My take

This is not Google being generous. This is a calculated response to fierce AI competition. By refusing to lock developers into their own models, Google ensures Android Studio remains the dominant platform. They are letting Claude and Codex do the heavy lifting while keeping developers firmly inside the Android ecosystem.

Original reporting: FoneArena. This is my rewrite and opinion.

More AI news for builders