OpenRig is an open-source TypeScript CLI and harness for assembling Claude Code, Codex, and related agents into a persistent, role-based team that you can boot from YAML and manage as one system.
OpenRig is an open-source system for building and running a network of AI coding agents. The README presents it as a way to turn separate terminal-based agent sessions into an organized, persistent team with shared context, owned work, and a lead agent that can coordinate specialists. It supports Claude Code and Codex together in the same rig, and the repo metadata also points to multi-agent orchestration, agent skills, CLI usage, tmux, and TypeScript.
The project is aimed at the problem of agent sprawl: instead of many disconnected terminal sessions, you get a repeatable setup for a team that keeps its context and responsibilities in one place. The README also emphasizes reducing friction around setup, permissions, and provider choice, so users can start with one repository and one useful change rather than managing ad hoc agent sessions. It is not presented as a general-purpose automation platform; it is specifically positioned around coding agents and team-style coordination.
At a high level, you define an agent team in YAML and launch it through the CLI. OpenRig then sets up the chosen provider harnesses, checks the required local environment, and applies workspace and permission settings so agents can operate under a team default. The README describes a lead-agent model: you tell that lead the outcome you want, and it coordinates specialist agents, keeps work and context persistent, and returns decisions or results for your attention. Some setup steps can also write provider hooks and trust settings, and there is an optional mode with broader access and non-interruptive behavior for certain workshop-style rigs.
It is gaining attention because it sits directly in the fast-moving AI coding-agent space and offers a practical way to run multiple popular agents together. The repository already has strong star growth, and the README targets current workflows with Claude Code, Codex, CLI-based setup, tmux, and multi-agent coordination. Its appeal is also likely helped by the fact that it promises a more structured, persistent alternative to juggling isolated agent sessions.
The README does not name direct competitors, but it implies alternatives such as using Claude Code or Codex on their own in separate terminals, or managing agents manually without a shared rig. A related option mentioned in the README is a workshop bundle with broader access and non-interruptive mode, which seems to be a different preset within the OpenRig ecosystem rather than a separate product. Based on the provided material, no other comparable tools are explicitly identified.
AI-explained · grounded in each repo's README