I shipped a QA harness for Claude Code. Instead of clicking through flows by hand, it reads your code diffs, identifies the affected UI flows, and tests them in real browser
Plus after each run you get screen recordings with console logs, network requests, HARs, and Playwright traces so you can inspect exactly what the agent did :)
https://github.com/wizenheimer/canary
P.S. I attempted to do a Show HN but got flagged for some reason