Local browser automation is mostly setup — install Chromium, keep the driver in step, keep the window off your screen. Running it in BrowserCat's cloud removes all of that and returns the same things: pages, screenshots and console output.
An MCP server for BrowserCat's cloud browser service. The browser runs on BrowserCat's infrastructure, so nothing is installed locally, and console output and screenshots come back as resources at `console://logs` and `screenshot://<name>`.
- Navigation to any URL
- Screenshots of the whole page or of a single element by CSS selector, at a width and height you set, or at the server's own default size
- Clicks and hovers on elements by CSS selector
- Input fields filled and dropdown options selected
- JavaScript executed in the browser console
- Browser console output readable as a resource, so a page error is visible without squinting at a screenshot
A BrowserCat API key as `BROWSERCAT_API_KEY` — free keys are available at browsercat.xyz/mcp. Run it with `npx -y @browsercatco/mcp-server`. Every interaction is selector-driven: the calls take CSS selectors, so a page that needs a click by sight rather than by selector is outside what this does.
One command plus a key — npx -y @browsercatco/mcp-server, then supply credentials