Hey Product Hunt 👋 A few months ago I realized I wasn't writing most of my code anymore, I was reviewing what agents wrote. But my tools still assumed one agent, one terminal, one branch. Every idea waited in line behind the last one. There are a couple of apps that let you run agents in parallel, and I loved the idea. But they were Mac only, didn't integrate the harnesses I wanted, and had limits on how many agents I could spin up. So I built the version I wanted to exist. Maestro runs on Mac, Windows and Linux. It works with whatever agent CLI you already pay for, Claude Code, Codex, Cursor, OpenCode or Kimi, through the logins those tools already have. Maestro never touches an API key. The parts I'm proudest of came from living in it every day. The preview tab is an embedded browser the agent can actually drive. It navigates, takes screenshots, reads the console and clicks buttons. You can scribble annotations on the page and send them back as feedback. Two agents can share one workspace, so one implements while another reviews. You can also leave inline comments on the diff and they land straight in the agent's next prompt. Merging is gated on approval, green CI, resolved comments and finished todos, because an agent saying it's done was never a great bar. And if your laptop is too small for the job, a project can live on a remote machine over SSH and everything runs there, terminals included. Today, I'm consistently using Maestro to build itself. If you're juggling agents across a pile of terminal tabs today, I'd love to hear what's missing. And if you're coming from a similar Mac only tool, there's a one click importer for your workspaces.

Maestro

Pilot a fleet of 20+ coding agents with zero collisions

Pilot a fleet of 20+ coding agents with zero collisions