Type: url
Original: https://ai.google.dev/gemini-api/docs/interactions/computer-use
Status: generated
Summary
Gemini API reference documentation for Computer Use, an agent tool that lets a model “see” a computer screen via screenshots and “act” by generating UI actions such as clicks, keystrokes, and scrolling for browser, mobile, or desktop environments, covering the request/response agent loop, supported actions per environment, configurable safety policies and confirmation requirements, opt-in prompt-injection screenshot scanning, and recommended safety practices such as sandboxing and human-in-the-loop confirmation.