🔥 Hot Repo: Boot a Claude-Codex Dev Team With One Command

One-liner — OpenRig is a multi-agent harness that lets you define Claude Code and Codex teams in YAML and run them as a single, coordinated system from one command. - Repo: mvschwarz/openrighttps://github.com/mvschwarz/openrig - Stars: ⭐ 2,744 +737 today - Language: TypeScript - License: Apache-2.0 --- What It Does OpenRig manages the system AI coding agents form when you run them together — not the agents themselves. You define topologies in YAML called RigSpecs, boot the whole team with rig up, and get a TUI showing every seat, its model, context usage, and runtime state. Agents communicate via rig send, rig broadcast, and shared chatrooms; the human steps in only for decisions. Why It's Blowing Up The multi-agent coding wave hit critical mass in 2026. Developers found themselves juggling four terminal windows — a Claude Code session here, a Codex session there — with no coordination layer on top. OpenRig fills that gap, and the timing matches a surge in "agentmaxxing": running multiple AI coding agents from different vendors in parallel across isolated git worktrees. A second driver is the "system-of-systems" pattern the community has been describing. Senior devs report running a dev pod implementor + QA + design alongside an adversarial review pod where Claude and Codex cross-check each other's work. OpenRig gives that pattern a proper harness instead of bash glue. The project was pushed today September 30, 2026 and added 737 stars in 24 hours — outsized for a developer-infrastructure tool requiring tmux and Node.js 22. The HN Show thread surfaced the core pitch clearly: it's designed to be driven by your agent, not by you typing commands by hand. That framing — the human as coordinator, not typist — resonates with how developers are actually working in 2026. Key Features - RigSpec YAML — define pods, edges, and seats declaratively; OpenRig wires up the tmux sessions automatically - Mixed provider support — same rig can hold Claude Code seats and C