Google Android Studio Rabbit 2 adds Bring Your Own Agent for 40+ coding agents
Google stabilized Bring Your Own Agent in Android Studio Rabbit 2, letting developers run more than forty coding agents inside the IDE through the open Agent Client Protocol, including Anthropic’s Claude Agent and OpenAI’s Codex from a built-in marketplace. Coverage tied to Android Developer Summit announcements also highlights deeper hooks into devices, performance profilers, and production crash reports across Android Studio and the terminal Android CLI, positioning agents to investigate bugs, test changes, and verify fixes without mandating Gemini. Google reported internal tests where new terminal tools accelerated UI-task completion by seventy-seven percent while using up to three times fewer tokens, metrics that remain company-run rather than independently validated. For platform operators, the shift signals openness over single-vendor lock-in, yet adoption and real-world productivity gains are still unproven in the excerpts available here.
Google Android Studio Rabbit 2 adds Bring Your Own Agent for 40+ coding agents
Google launched a Bring Your Own Agent feature in Android Studio Rabbit 2 so developers can run more than 40 coding agents in the IDE via the Agent Client Protocol, including Anthropic Claude Agent and OpenAI Codex from a built-in marketplace. Google reported internal tests where terminal tools helped agents finish UI tasks 77 percent faster while using up to three times fewer tokens.
Key takeaway
Google is opening Android Studio to rival coding agents via a stable BYOA layer, but public excerpts offer no proof those agents deliver at scale.
What happened
According to The Decoder and owned-news reporting, Google launched Bring Your Own Agent in Android Studio Rabbit 2 so developers can run more than forty coding agents in the IDE via the Agent Client Protocol, with Anthropic Claude Agent and OpenAI Codex available from a built-in marketplace.
The New Stack reports that Google announced at Android Developer Summit that agents such as Claude Code and OpenAI Codex gain access to Android devices, performance profilers, and production crash reports across Android Studio and the terminal Android CLI, and Google said internal terminal tooling helped agents finish UI tasks seventy-seven percent faster with up to three times fewer tokens.
Evidence
Android Studio Rabbit 2 supports more than forty coding agents through Bring Your Own Agent and the Agent Client Protocol.
The Decoder · attributed
Through a new "Bring Your Own Agent" feature, the IDE now supports more than 40 coding agents, including Anthropic's Claude Agent and OpenAI's Codex.
Google reported internal UI-task speed and token-use figures for new terminal tools.
The Decoder · attributed
Google reported internal tests where terminal tools helped agents finish UI tasks 77 percent faster while using up to three times fewer tokens.
Bring Your Own Agent is stable in Android Studio Rabbit 2 and spans Studio plus the Android CLI.
The New Stack AI · attributed
The Bring Your Own Agent feature supports over 40 agents using the open Agent Client Protocol and is now stable in Android Studio Rabbit 2, spanning both Android Studio and the terminal-based Android CLI.
Third-party agents can use devices, profilers, and production crash reports to debug and verify fixes.
The New Stack AI · attributed
Google is giving coding agents like Anthropic’s Claude Code and OpenAI’s Codex deeper access to its Android development tools, including devices, performance profilers and production crash reports
Why it matters
Android builders can plug preferred agent stacks into official device, profiler, and crash workflows instead of maintaining parallel integrations outside Google’s toolchain.
Limits and uncertainties
The New Stack analysis notes excerpts do not establish adoption rates, performance outcomes, or competitive impact.
Seventy-seven percent faster UI tasks and three-times-fewer-token claims come from Google’s internal tests, not independent benchmarks in the packet.
Several bundled feed items (Towards AI, Les Echos) provide only headlines or minimal excerpts unrelated to BYOA depth.
Practical implications
Evaluate Agent Client Protocol-compatible agents against your Android CLI and Studio workflows before standardizing on one vendor.
Treat profiler, device, and crash-report access as sensitive when routing third-party agents that may handle production telemetry.
What to watch
Whether Google publishes independent or developer-reported productivity data beyond the cited internal UI-task tests.
How quickly teams adopt non-Google agents in Rabbit 2 versus Google’s own Antigravity and Gemini offerings mentioned in related coverage.
