Hey Product Hunt 👋 We built this for a problem we kept hitting ourselves. AI writes our code now — Cursor, Claude Code, Copilot. But testing? That still meant writing Playwright/Selenium scripts, or quietly skipping it and hoping. If you're a non-developer or a "vibe coder" shipping apps with AI, that gap is brutal: you can build a whole app without reading code, but you can't confirm it actually works without it. Specnote closes that gap. You describe what you want to check in plain language — "sign up, log in, and make sure checkout works." An AI agent then drives a real browser through your app like a person would, clicking and typing step by step, and hands you a fix report pinpointing exactly where it broke. No test code, no selectors. It connects to your codebase over MCP, so it runs right alongside Cursor, Claude Code, and Copilot. Code access is read-only, and passwords / API keys are masked automatically. If you've ever shipped something built with AI and thought "…wait, does this still work?" — that's exactly who we made this for. I'll be around all day. Would love your honest feedback, and happy to answer anything

Specnote

You handed the coding to AI. Now hand it the testing, too.

You handed the coding to AI. Now hand it the testing, too.