Openrig

OpenRig coordinates coding agents as a persistent team. A user talks to one lead agent, which delegates to specialist agents, tracks their work and brings answers back, so that Claude Code, Codex and Pi can work together on projects that run for weeks or months. A team is defined in a rig spec that names each agent, its role, harness, model and tools. A single command starts every agent, each in its own tmux session, and another stops them together. Teams can mix harnesses, and the spec can set a model and reasoning effort for each agent. Agents message one another and share a work queue, and a delegation edge sets who gets alerted when work sent to an agent stalls. Specialists can be asked direct questions without creating a new task. Agents can be added to or removed from a running team, and agents already running in tmux can be discovered and adopted without a restart. Teams can also span more than one machine. The features page lists 46 features, each linked to documentation. It is aimed at developers running large builds who want durable, specialized agents instead of one long session. It is self hosted and needs Node 22 or 24, tmux and a signed in Claude Code or Codex. Setup is offered either through an agent skill or an install script that starts a local daemon. The site does not publish pricing.