Hey HN,

I’m a developer and 3D artist, and I wanted my AI (specifically Cursor and Claude Desktop) to have "hands and eyes" in my actual browser while I work.

I tried the official Chrome DevTools MCP, but it felt overkill for my workflow. It requires setting up remote debugging ports and is heavily geared toward performance profiling and deep-dive engineering. I just wanted something "generalized"—like ChatGPT Atlas or Comet—but for my own dev environment.

So, I built Apex Agent.

It’s a lightweight Chrome extension + a tiny Node.js bridge that lets any MCP-compatible AI control your browser session.

Why I built this vs. using the official MCP: Human-Centric Tools: While the official tool is for debugging, Apex is built for interaction. It has 69+ tools for clicking, typing, scrolling, and taking full-page screenshots. No Remote Debugging Mess: You don’t need to restart Chrome with special flags or mess with debugging ports. Just connect the extension and go. Control Your Active Session: It doesn't spawn a separate "headless" instance. It works with the tabs you already have open, which is way more useful for vibe coding or UI testing. Dev Workflow Ready: I optimized this specifically to work with Cursor. Now I can tell Cursor to "Go to my local dev site, find the submit button, and tell me if the console shows any errors" without leaving my editor.

I’m looking for feedback from the community—what tools are missing for your daily AI workflows?

GitHub: https://github.com/RTBRuhan/ApexAgent

Chrome Web Store: https://chromewebstore.google.com/detail/apex-agent/pmpkkbjd...