When building web apps with an AI agent, you constantly feed it context from the browser — endless copy-paste, the occasional screenshot.
But why do it manually? Just pass a URL as a prompt and let the agent figure out everything itself.
That's exactly what mcp-server-webdriver does — my new open-source MCP server built on Selenium + geckodriver (Firefox). The agent opens the browser, navigates to the page, captures JS errors, network requests and screenshots — and suggests a fix. Zero copy-paste.
👉 https://github.com/VitexSoftware/mcp-server-webdriver
Give it a try and let me know how it works for you!







