// Agent profile
mvschwarz/openrig is a multi-agent orchestration harness that manages coding agents as a coordinated team rather than isolated terminal sessions. Define your agent team in a YAML RigSpec — pods, members, edges, continuity policies — and boot the entire team with a single command. Claude Code and Codex run side by side in managed tmux sessions with cross-agent messaging, broadcast channels, and shared chatrooms. OpenRig ships a terminal UI dashboard for monitoring team state and topology, MCP tools that let agents self-manage their own topology at runtime, and snapshot/restore for capturing and resuming full team states. At 3,500+ GitHub stars and actively trending, OpenRig represents the emerging multi-agent harness category where the unit of work shifts from a single agent session to a coordinated team.
OpenRig is a multi-agent harness from mvschwarz that turns scattered terminal agent sessions into an organized, persistent team. Where most agent tooling focuses on making a single agent more capable, OpenRig focuses on making multiple agents work together — defining team structure, managing communication channels, and maintaining team state across sessions. The project has 3,500+ GitHub stars and is trending on GitHub.
The central abstraction is the RigSpec, a YAML file that declares the team topology: which agents run, how they are grouped into pods, what communication edges connect them, and what continuity policies govern session handoff. A single rig up command boots the entire team. Under the hood, each agent runs in a tmux session, which means users can attach directly to any agent’s terminal for inspection or manual intervention.
Cross-agent communication goes beyond simple message passing. OpenRig provides point-to-point messaging, broadcast to all team members, and persistent chatrooms where agents can coordinate asynchronously. The MCP integration lets agents modify their own topology at runtime — adding new team members, removing finished ones, or restructuring pods based on task requirements.
The snapshot/restore system captures full team state — every agent’s session, the topology graph, communication history — and can resume it later. This solves the cold-start problem for multi-agent workflows: instead of rebuilding context from scratch, restore a checkpoint and continue where the team left off.
OpenRig is an orchestration layer, not an agent runtime — it manages existing agents (Claude Code, Codex) rather than replacing them. The tmux-based architecture means it runs locally on macOS or Linux. The project is actively evolving (currently at v0.5.x) and the API surface should be expected to change. Team topologies add coordination overhead that only pays off when the task genuinely benefits from multiple specialized agents rather than a single capable one.
Developers and teams who regularly run multiple coding agents and want structure around how those agents coordinate. Anyone who has found themselves managing three terminal tabs with different agents and wished for a unified control plane. Teams experimenting with multi-agent software development workflows who want declarative, reproducible team configurations.
Builder.io's open-source framework for building agent-native applications — shared actions, SQL-backed state, identity, tools, skills, jobs, observability, and UI surfaces that all work together.
CLI orchestration tool that installs and configures platform connectors for 16+ social and web sources — Twitter/X, Reddit, YouTube, GitHub, Bilibili, and more. Zero API fees.
Persistent memory layer for AI coding agents — benchmark-backed (95.2% on LongMemEval-S), 92% fewer tokens per session vs full-context pasting, zero manual memory.add() calls.