Hey Product Hunt 👋 I'm Sebastian, and I built Cowboy. Every team has review rules: "big PRs need two reviewers", "migrations need an owner", "agent PRs get a closer look". Almost nobody checks whether those rules actually catch problems, or just cost reviewer time. Cowboy sorts each pull request into easy, normal or careful with rules you can read, flags shortcuts like skipped tests or silenced warnings, and tracks what happens after merge: reverts, broken main, follow-up fixes. Then it tests your rules against simple alternatives on your own history. We ran that test on our own default rules first, on about 28,000 merged PRs from 11 public repos. They didn't win everywhere. That's the point: no rule set works for every team, so Cowboy measures yours instead of asking you to trust it. Every result, including what failed, is on the Evidence page. It's free and open source, and it runs on your machine. Your code never leaves. You can rate your last 90 days on day one.

Cowboy

Your local PR partner: sort PRs, test your review rules

Your local PR partner: sort PRs, test your review rules