At Google Cloud Summit London in early July, Google revealed significant new capabilities in Firebase Studio that promise to enhance agentic cloud-based development: an autonomous Agent mode, native support for Model Context Protocol (MCP), and Gemini CLI integration. These updates aim to streamline agentic AI development by making AI agents more independent, better informed, and seamlessly embedded in developer workflows.
The new autonomous Agent mode allows Gemini to work independently on behalf of developers. In this enhanced mode, Gemini can independently reason, take actions such as running terminal commands, and generate entire applications without constant developer oversight. It can also add features to existing applications, like refactoring components, writing tests, and fixing errors. For critical actions carrying irreversible consequences, such as deleting files, the system will prompt developers for explicit permission before proceeding.
The autonomous Agent mode represents a major leap forward from Firebase Studio’s existing interaction capabilities. This new mode builds upon the platform’s Ask mode, which enables conversational interaction with Gemini, and the prior Agent mode, which allows Gemini to propose application changes for developer approval. The key innovation is Gemini’s ability to autonomously reason and work independently, significantly accelerating development tasks.
Source: Google blog post
Firebase Studio also now offers native support for MCP, allowing developers to leverage MCP servers to extend and personalize their development workflow. The protocol is a standardized way to connect AI models to different data sources and tools. For example, while writing code or implementing new features, developers can use natural language to interact with a Context7 MCP server to look up and study specific APIs, or interact with a Postgres MCP server to learn about existing database tables and their column types. This native integration streamlines access to crucial development context, reducing friction and accelerating tasks.
A third major update integrates Gemini CLI directly into Firebase Studio. Google launched their free and open-source AI agent called Gemini CLI in June 2025. With this integration, developers who primarily use a command-line interface (CLI) to control their development workflow can now use Gemini’s capability directly within the Firebase Studio. This eliminates the need to switch contexts to a separate chat window, allowing for a more seamless and integrated AI-assisted development experience right where they work.
Firebase Studio is Google’s agentic cloud-based development environment designed to help build and ship full-stack AI applications. These three new capabilities collectively represent Google’s push to make AI-assisted development more autonomous, contextually aware, and integrated into existing developer workflows.
This expanding landscape of AI-powered development tools includes other notable platforms such as Lovable, which focuses on generating websites and full-stack applications from natural language descriptions. Similarly, Bolt.new is an AI-powered web application builder that simplifies the creation of web applications. Replit, another online AI-powered platform, enables users to quickly transform ideas to web applications and deploy them directly from the browser, eliminating the need to wrestle with complex development environments. Like Firebase Studio, these tools aim to streamline and accelerate the software development lifecycle through autonomous and context-aware AI agents.