Computer use is now in Claude Code
Claude can now open apps, click UI elements, and test code directly on your desktop.
Anthropic has unveiled a groundbreaking new capability for its Claude AI assistant called Claude Code, currently available as a research preview for subscribers on Pro and Max plans. This feature represents a significant leap from conversational AI to an interactive agent, granting Claude the ability to directly control a user's computer. Through the command line interface (CLI), Claude can now execute commands to open applications, manipulate graphical user interfaces by clicking buttons and navigating menus, and run tests on the software or code it has just written or modified.
This transforms Claude from a coding assistant that suggests snippets into an active participant in the software development lifecycle. Developers can instruct Claude to build a feature, then watch as it autonomously opens the relevant IDE or code editor, implements the changes, launches the application, and performs functional tests to verify the results. This closed-loop automation promises to drastically accelerate prototyping, debugging, and quality assurance processes by removing manual steps between AI-generated code and real-world execution.
- Claude Code is a research preview feature for Pro/Max users, enabling direct computer control via CLI.
- The AI can open applications, click UI elements, and test its own code output autonomously.
- This shifts Claude from a conversational assistant to an actionable agent in the development workflow.
Why It Matters
This turns AI from a coding co-pilot into an autonomous engineering agent, potentially automating entire development and QA workflows.