Google has introduced a new Android command-line interface built specifically for AI agents, claiming a 70 percent cut in token usage and three times reduction in task completion time.
The primary Android development environment remains Android Studio, which has built-in support for AI agents, but the new CLI means that agents working outside Android Studio can more easily build Android applications. The CLI is not a replacement for Android Studio, and applications built with the CLI can also be opened in the IDE (integrated development environment).
“You can start a prototype quickly with an agent using Android CLI and then open the project in Android Studio to fine-tune your UI,” states the introductory post.
The CLI itself is not powered by AI and is also suitable for use by scripts and other automation tools, or by developers who prefer working in a code editor rather than a full IDE.
Installing the Android CLI, available for Apple silicon, AMD64 Linux and AMD64 Windows, enables the android command, with arguments for creating applications from templates, installing and managing the Android SDK and device emulators, and finding and listing Android skills, these being instruction files which assist agents to perform specific tasks. There is also a “describe” argument, which analyzes a project and generates descriptive metadata; and a “docs” argument, which searches and fetches documentation from the Android knowledge base.
Android skills are available in a GitHub repository. Currently only 7 skills are listed but more are likely to follow.
The terms of service reveal that Google collects Android CLI usage data by default, to “help improve the tool.” Commands can be excluded from data collection by adding the –no-metric flag.
Early reaction to the Android CLI is mixed, with some developers finding it too limited. “All it offers is some wrapper around the basic Android setup command that LLMs are already good at,” said one.
The CLI is likely to improve though, and more skills and templates will become available.
Traditional IDEs are optimized for use by humans rather than AI agents, and Google is not the only vendor thinking about how AI will reshape development tools. Microsoft talked of a “fundamental shift in how we think about IDEs,” in reference to a new agentic Visual Studio feature; and JetBrains has previewed Central as a system for agentic software development.
One of the functions of an IDE is to make it easy to see and navigate all the code that forms a software project. Having AI agents work with the command line instead helps them to work more efficiently, but may further distance software development from the developer, as code is generated and compiled behind the scenes.
