Android Studio Embraces Flexibility with "Bring Your Own Agent" Support for Claude, Codex, and Antigravity

Artificial intelligence-powered developer tools have fundamentally transformed engineering productivity across the software industry, evolving from simple autocompletion helpers into essential productivity multipliers. Modern development teams routinely rely on specialized AI coding agents, custom enterprise harnesses, and autonomous tools to streamline their workflows, reduce repetitive tasks, and accelerate time-to-market. Recognizing that no single AI model or workflow suits every developer, the ecosystem is shifting toward greater openness. Building upon last year’s milestone that allowed Android Studio to integrate with any remote AI model, the platform is now taking the next logical step by introducing native support for a wide variety of coding agents.

With the introduction of the new "Bring Your Own Agent" (BYOA) feature, developers can seamlessly integrate their preferred coding assistants directly into Android Studio. This capability bridges the gap between external AI models and the deeply optimized infrastructure and tool support native to the official Android integrated development environment. Rather than forcing development teams into a rigid, single-provider ecosystem, BYOA empowers engineers to pair their favorite external coding agents—including Anthropic’s Claude Agent, OpenAI’s Codex, and Google’s Antigravity—with IDE-native intelligence designed specifically for mobile application development.

Your agents, your AI plan

The introduction of BYOA is designed to give developers absolute control over their tooling stack and subscription plans, pairing individual agent preferences with deep, IDE-native intelligence. This architectural approach aims to make AI-assisted workflows significantly faster, more accurate, and more cost-effective for individual developers and large engineering organizations alike. By running inside the Android Studio environment, these external agents gain direct contextual awareness of the project, allowing them to leverage optimized infrastructure without sacrificing the specialized reasoning capabilities of specific model families.

This flexibility acknowledges a growing reality in modern software engineering: developers often have strong preferences regarding which foundational models and agent architectures they trust for specific tasks. Whether an engineer prefers the conversational nuance and architectural planning of Claude, the code generation strengths of Codex, or the specialized Google integrations found in Antigravity, the BYOA framework ensures they can remain within their preferred development environment while utilizing these tools. Available starting in the latest Android Studio Canary releases, the feature removes historical friction points that previously required developers to constantly switch between external chat interfaces and their local codebases.

Powerful, capable, and reliable coding agents

Modern coding agents do far more than suggest single lines of code; they are capable of planning and executing complete, complex technical workflows directly within a developer’s local environment. This level of autonomy unlocks powerful use cases that were previously impossible or required tedious manual coordination. Agents can analyze entire project structures, orchestrate multi-file refactoring operations, write comprehensive test suites, and diagnose subtle runtime issues by interacting directly with build logs and debugging tools.

When running inside Android Studio, these external agents do not operate in a vacuum. They benefit directly from specialized Android skills and the comprehensive Android Knowledge Base. This integration ensures that regardless of which underlying model or agent provider a developer chooses, the resulting code and architectural suggestions have access to the latest Android best practices, Jetpack library updates, and modern architectural guidelines. Furthermore, for those interested in the academic and empirical side of model capabilities, Google highlights recent updates to Android Bench, which provide deeper insights into how advanced agent architectures handle long-horizon, multi-step mobile development tasks.

Using Gemini with Google Antigravity

While the introduction of BYOA opens the door to third-party models, Google remains deeply committed to providing a robust, native Gemini experience within Android Studio. Many developers have grown accustomed to using Gemini directly through the built-in agent, and this familiar experience will continue to be fully supported. However, for developers looking to maximize their Gemini capabilities, Google recommends selecting the specialized Google Antigravity agent.

Choosing the Google Antigravity agent grants developers direct access to the latest frontier Gemini models, such as Gemini Flash 3.8, alongside significantly increased AI usage quotas. Developers who maintain an active Google AI Pro or Ultra plan can easily log into the Antigravity agent within Android Studio to immediately take advantage of their existing subscription benefits. Alternatively, teams and individuals who prefer usage-based billing can opt to pay per token rates by simply supplying their own Gemini API key, ensuring complete flexibility in how their AI usage is funded and managed.

Selecting the Google Antigravity agent

Enterprise environments require stringent adherence to data privacy, security governance, and compliance standards. For organizations utilizing Gemini Enterprise, the transition to agentic workflows is fully supported without compromising corporate security postures. Enterprise developers can choose to continue using the built-in default agent or transition to the Antigravity agent depending on their specific organizational policies and workflow preferences. In either scenario, corporate users continue to benefit from the enterprise-grade security and strict data privacy guarantees provided by Google Cloud, ensuring that proprietary source code and internal intellectual property remain protected while leveraging advanced AI capabilities.

The Antigravity agent registry is designed to handle diverse organizational authentication methods seamlessly. Developers can log in using standard Google accounts, apply enterprise Gemini licenses, connect to dedicated enterprise agent platforms, or authenticate via API keys. This modular authentication framework ensures that whether an individual developer is experimenting on a personal project over the weekend or a multi-thousand-person engineering organization is deploying AI tools across global teams, the integration layer remains secure, compliant, and adaptable to existing enterprise IT infrastructure.

Get started

Support for the Bring Your Own Agent feature is currently rolling out in preview, debuting in the canary release of Android Studio Rabbit 2. This initial rollout supports commonly used agents including Google Antigravity, Claude Agent, and Codex, accommodating both consumer and enterprise AI plans depending on the specific policies and offerings of each agent provider. Developers interested in testing these capabilities can download the latest canary build, navigate the configuration menus to connect their preferred agent, and begin experimenting with autonomous workflows in their current mobile projects.

As the Android development community moves deeper into the era of agentic tooling, platform maintainers emphasize that user feedback remains essential for refining the IDE experience. Developers who encounter unexpected behavior, bugs, or performance friction while using external agents with BYOA are encouraged to file detailed issue reports through the official Android Studio bug tracking channels.