Google’s Firebase Studio, an AI-powered development workspace hosted in the Google Cloud, has added new AI-assisted development tools including an autonomous agent mode, foundational support for MCP (Model Context Protocol), and Gemini CLI integration.
The improvements were announced July 10. Three distinct modes now are available for interacting with Google’s Gemini AI model:
- Ask – a mode used for discussion and planning with Gemini and geared toward planning code, brainstorming, and discussing complex problems collaboratively.
- Agent – a mode in which Gemini can propose changes to an app, but the proposed changes must be approved by the developer before any files are modified.
- Agent (Auto run) – a mode in which Gemini can autonomously reason, generate entire apps, and add features to existing apps without explicit developer approval. However, developer permission is required before deleting files, running terminal commands, or using external tools.
The Agent and Agent (Auto-run) modes can leverage personalized guidance from a variety of project-level rule files to ensure Gemini adheres to specific design patterns and preferences, Google said. Gemini automatically detects and loads instructions from files like .idx/airules.md, GEMINI.md, or .cursorrules at the project root.
