Computer Use
Browser, desktop and Chrome control in one install
Computer Use gathers the three ways Whilo drives a computer. Browser automation runs the tabs Whilo manages: open and close tabs, read a page snapshot, click, type, press keys, fill 1 to 16 ordinary main-frame fields from one fresh snapshot, and run many steps in a single scripted call. Each action passes the tab queue, so a person who takes a tab stops the agent until it is released, and a password, one-time code, CAPTCHA or payment goes through secure input or a takeover. Computer use reaches the rest of the desktop through cua-driver, the MIT driver Whilo packages and pins: it reads the screen and the accessibility tree and posts input, and because Whilo spawns it directly, macOS grants Screen Recording and Accessibility to Whilo rather than to a helper. Both of those run inside Whilo already. This listing adds the third: Chrome over the DevTools Protocol, through the ChromeDevTools server, which drives a Chrome window of its own with navigate, snapshot, click, type, network and performance reads, and screenshots.
How to connect Computer Use to Whilo
Open Computer Use in Whilo
Press Open in Whilo on this page, or find Computer Use in the app's Apps tab. Get the app first if it is not on this computer.
Give an agent a task
Whilo asks for no sign-in or key for Computer Use. Ask an agent for work that needs Computer Use, such as one of the example prompts below. The agent uses the plugin's tools to do it.
Example prompts3
Computer UseOpen the pricing page, read what each plan includes, and tell me which one fits a team of two.
Apps1
Chrome DevTools
A Chrome window of its own, driven over the DevTools Protocol: navigate, snapshot the page, click, type, read network and performance, and take screenshots.
Information4
- Developer
- Whilo
- Category
- Automation
- Version
- 1.0.0
- Website
- github.com/trycua/cua