Computer Use
Claude can see your screen and operate your computer — clicking, typing, navigating browsers, and running desktop apps. This section covers the Claude Computer Use API, browser automation, and using Claude Code as a local automation agent.
In This Section
What is Claude Computer Use
How Claude sees screenshots and sends mouse and keyboard actions — capabilities and current limitations.
How Computer Use Works
The vision and action loop: screenshot, decide, click or type, observe, repeat.
Browser Automation with Claude
Claude navigating web pages, filling forms, scraping data, and clicking through multi-step web flows.
Desktop & GUI Automation
Claude controlling desktop applications — file dialogs, menus, windows, and system-level interactions.
Claude Code as a Local Agent
Using Claude Code CLI to read and write files, run shell commands, and automate local dev tasks.
Building a Computer Use Pipeline
Anthropic SDK and computer use tools — an end-to-end working example of a computer use pipeline.
Computer Use Safety & Sandboxing
What Claude will and will not do, sandboxed VMs, and human approval gates for risky actions.