AI
AI
AI
Google LLC’s Android team is introducing new ways to build high-quality software for its mobile platform with artificial intelligence agents.
Today, the company provided two new tool suites, including a new Android command-line interface with skills designed to enable agents to build cleanly and a knowledge base filled with official documentation.
These two new resources complement each other. The CLI, or command-line interface, is a lightweight way for developers to interact with AI coding agents, eschewing graphical windows and going directly to the terminal, which relies entirely on text and keyboard input rather than buttons and icons.
Google said it rebuilt this interface for Android development, providing developers with a way to surface commands for environment setup, project creation and device management right at their fingertips. This way, they can get right down to the nuts and bolts without the distractions of a graphical interface.
The company also said the thinner client reduces the number of tokens used by large language models during development by more than 70% and tasks were completed more than three times faster than standard toolsets. Although they did not explain exactly where these optimizations came from.
New capabilities baked into the updated CLI include broader software development kit management with “android sdk install” for downloading specific components, access to “android create” to generate new projects from official templates and rapid virtual device creation.
Normal documentation is designed for humans to read, learn and understand examples so that they can build and design software; however, this works against LLMs and AI agents. AI agents require rigorous, actionable instructions to execute step-by-step tasks, sometimes with static examples and edge cases laid out.
Although AI agents can parse human-readable documentation, it slows them down and causes confusion. To avoid this, the Android team is launching the Android Skills GitHub Repository.
Skills are modular, markdown-based instruction sets that include technical specifications for tasks designed to trigger automatically when a prompt matches specific criteria. Developers who use agentic systems typically know them as SKILL.md files and accompanying metadata. They simplify routines and common prompts.
The company said Android skills cover some of the most typical workflows that Android developers and LLMs struggle with across the ecosystem and follow best practices. In the initial release, the Android team included skills to handle Navigation 3, support for Android Grade Plugin 9, XML-to-Compose, R8 config analysis and more.
Android CLI is available in preview today, along with its growing set of skills, on Google’s Android developer tools portal.
Finally, Google said it’s launching a fully featured documentation portal known as the Android Knowledge Base, now accessible through the “android docs” command and in the latest version of Android Studio.
It is a specialized data source that allows coding agents to quickly search and fetch the latest authoritative developer guidelines to use as context. The company explained that the Knowledge Base is constantly updated, providing agents with a foundation to ground responses and remains up to date with recent information from Android development. This ensures LLMs keep on top of guidance even after the LLM training cutoff date.
Now, more than ever, developers face a challenge where code can suffer from LLMs churning or producing bad code based on outdated knowledge. This requires either manual cleanup, further iterations with the AI chat or another agentic layer to fix the initial run, wasting even more AI tokens. If an LLM can get as much clean code written in the first pass, following best practices, within a company’s own stylesheet, it makes it more maintainable over time.
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.