Google is giving third-party coding agents such as Anthropic's Claude Code and OpenAI's Codex deeper access to its Android development tools, including devices, performance profilers and production crash reports, letting agents investigate bugs, test changes and verify their own fixes. The announcements were made Wednesday at the Android Dev Summit.
Google's push spans Android Studio and the terminal-based Android CLI, with the company opening both to outside agents rather than steering developers toward its own Gemini assistant. Matthew McCullough, vice president of product for Android, has emphasized giving developers the freedom to choose their own models and coding agents. In a Wednesday post, Adarsh Fernando, senior product manager for Android developer tools, described the goal as giving developers and their agents "the tools and knowledge to build high-quality Android apps faster."
The company's Bring Your Own Agent feature now supports more than 40 coding agents that speak the open Agent Client Protocol, among them Anthropic's Claude Agent, OpenAI's Codex and Google's own Antigravity. Google says the feature is now stable in Wednesday's release of Android Studio Rabbit 2. Developers can install several agents and switch between them without leaving the IDE, and connected agents inherit the IDE's own understanding of a project, including its structure, build configuration, symbols and tools such as Compose Preview, the emulator and Logcat.
A new Change Review feature scans an agent's pending changes for common issues and can walk a developer through a large set of modifications before anything is applied. Google is also pushing the same reach beyond the IDE with official Android CLI plugins for Claude Code and Codex covering project creation, emulator management, UI inspection and Android documentation.
The new android preview command lets coding agents render Jetpack Compose previews directly in the terminal without launching an IDE or emulator. Google says agents using it completed UI tasks 77% faster while using up to three times fewer tokens, though it shared little about how those results were measured. The CLI also adds android device remote, giving agents access to physical devices in Google's lab for testing.
Google is also giving agents more ways to investigate problems after an app is built, including performance issues, production crashes and memory leaks. In one demo, Claude Code worked with LeakCanary, an open-source leak-detection library, to check critical user journeys for memory leaks using a single instruction. Google says the tooling helped capable models complete such investigations nearly twice as fast while using up to 25% fewer tokens. A companion android signals command pipes Android vitals and Firebase Crashlytics data into the same terminal workflow, giving agents a direct line to production crashes, ANRs and memory problems.
Google is also introducing Android Bench 2.0 to evaluate how models and agent harnesses perform on more complex Android development tasks, with assignments ranging from building an app from mockups to migrating an existing interface. The benchmark tests whether completed apps actually run and measures completion time and cost.
The article notes a record that counsels patience: autonomous bug-finding can outrun autonomous bug-fixing, citing an Anthropic scanner that flagged some 29,000 possible vulnerabilities in open-source code while only a few hundred were ever fixed. It also points to developers who watched Claude Code land a Linux kernel fix only for reviewers to catch a race condition, underscoring that the new tools do not retire the human from the loop.
Analysis
Why This Matters
- Android developers can now use a wide range of third-party coding agents inside Google's official tooling, rather than being limited to Google's own assistants.
- Agents gain direct access to production crash reports, performance data and lab devices, which could change how bugs are found, diagnosed and fixed.
- Android Bench 2.0 gives developers and companies a way to compare models and agent harnesses on realistic Android tasks, including completion time and cost.
Background
Coding agents are AI assistants that can write, edit and test code in a developer's environment. Android Studio is Google's official integrated development environment for building Android apps, and Google has been adding its own AI features to it. By opening the same tools to outside agents through the Agent Client Protocol, Google is positioning its platform as model-agnostic, letting developers bring whichever agent they prefer.
Key Perspectives
Google: says the goal is giving developers and their agents "the tools and knowledge to build high-quality Android apps faster," and its vice president of product has emphasized giving developers freedom to choose their own models and agents.
The New Stack: cautions that autonomous bug-finding can outrun autonomous bug-fixing, citing an Anthropic scanner that flagged some 29,000 possible vulnerabilities in open-source code while only a few hundred were fixed, and noting a Linux kernel fix that reviewers had to correct.
What to Watch
- Whether Google's claimed speed and token reductions for terminal previews and leak investigations hold up in independent testing.
- How quickly developers adopt third-party agents inside Android Studio and whether the Change Review feature meaningfully reduces bad agent output.
- Android Bench 2.0 results as models and agent harnesses are evaluated on complex, multi-day Android development tasks.
Checked
12 claims in this story checked against 1 source: 12 supported.
Show each claim
- SupportedGoogle is opening Android Studio to third-party coding agents such as Claude Code and Codex.The New Stack: “Google is giving coding agents like Anthropic’s Claude Code and OpenAI’s Codex deeper access to its Android development tools”
- SupportedBring Your Own Agent is stable, and Google has announced CLI plugins, terminal previews and an agent-performance benchmark.The New Stack: “Google says the feature is now stable in Wednesday’s release of Android Studio Rabbit 2.”
- SupportedAgents can access Android development tools including devices, performance profilers and production crash reports to investigate bugs, test changes and verify fixes.The New Stack: “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; now, agents can now investigate bugs, test changes, and verify their own fixes.”
- SupportedThe announcements were made Wednesday at Android Dev Summit.The New Stack: “The updates were announced Wednesday at Android Dev Summit .”
- SupportedAdarsh Fernando said the goal was to give developers and their agents “the tools and knowledge to build high-quality Android apps faster.”The New Stack: “In a Wednesday post , Adarsh Fernando , senior product manager for Android developer tools, describes the goal as giving developers and their agents “the tools and knowledge to build high-quality Android apps faster.””
- SupportedBring Your Own Agent supports more than 40 coding agents using the Agent Client Protocol, including Claude Agent, Codex and Antigravity.The New Stack: “The company’s Bring Your Own Agent feature now supports more than 40 coding agents that speak the open Agent Client Protocol , among them Anthropic’s Claude Agent, OpenAI’s Codex, and Google’s own Antigravity”
- SupportedGoogle says Bring Your Own Agent is stable in the Android Studio Rabbit 2 release.The New Stack: “Google says the feature is now stable in Wednesday’s release of Android Studio Rabbit 2.”
- SupportedGoogle announced official Android CLI plugins for Claude Code and Codex, covering project creation, emulator management, UI inspection and Android documentation.The New Stack: “Google is pushing the same reach beyond the IDE with official Android CLI plugins for Claude Code, and Codex that cover project creation, emulator management, UI inspection, and its Android documentation.”
- SupportedThe android preview command lets agents render Jetpack Compose previews in the terminal without launching an IDE or emulator.The New Stack: “the new android preview command that lets coding agents render Jetpack Compose previews directly in the terminal without launching an IDE or emulator.”
- SupportedGoogle says agents using terminal previews completed UI tasks 77% faster while using up to three times fewer tokens.The New Stack: “The company says agents using it completed UI tasks 77% faster while using up to three times fewer tokens, though it shared little about how it measured those results.”
- SupportedIn a demo, Claude Code used LeakCanary to check critical user journeys for memory leaks with a single instruction.The New Stack: “In one demo, Claude Code worked with LeakCanary, an open source leak-detection library, to check critical user journeys for memory leaks using a single instruction from the developer.”
- SupportedAndroid Bench 2.0 evaluates models and agent harnesses on complex Android tasks, tests whether apps run, and measures completion time and cost.The New Stack: “Google is also introducing Android Bench 2.0 to evaluate how models and agent harnesses perform on more complex Android development tasks. It tests whether completed apps actually run”
