Framer AI AgentsDesign and publish professional sites with AI
Promoted
Launching today
Noru Flow
Let your coding agent see what's on your screen
1 follower
Let your coding agent see what's on your screen
1 follower
Hands free: you never take the screenshot. Your agent asks over a local MCP server and gets one fresh capture plus the on-screen text, mid-task, without you cropping or pasting anything. Playwright already covers the DOM. Noru covers what it can't: native apps, simulators, system UI, final rendered pixels, and work spanning several apps. Nothing is uploaded. Captures stay on your Mac, Noru observes only, and every observation is acknowledged on screen first.
Hey Product Hunt, Glendon here.
I built Noru because I was tired of the screenshot loop. Every time my coding agent needed to see a native app, a simulator, or a layout bug that doesn't exist in the DOM, I'd stop, screenshot, crop, paste, and re-explain. The agent was capable. It just couldn't see.
Two things worth knowing up front.
It's hands free. You don't take the screenshot. Your agent asks over a local MCP server, mid-task, and Noru hands back one fresh capture plus the on-screen text. You aren't in the loop at all.
It fills a gap rather than replacing anything. If your agent is working on a web page, Playwright's structured browser state is the better tool, and Noru's own docs say exactly that. Noru is for what the DOM can't describe: native apps, simulators, system UI, final rendered pixels, and workflows that span several apps at once.
Nothing is uploaded. Captures stay on your Mac, Noru observes only (it never clicks or types), and every observation is acknowledged on screen (pop up) before the agent receives it.