OpenRig: Run Claude Code and Codex as a Multi-Agent Dev Team
By Matt Cullerton, founder of Mavric and Apso
OpenRig is an open-source tool that runs Claude Code, Codex and other coding agents as a persistent team, with roles, shared context and agents that message each other. A TikTok pitching it as a way to "Turn Claude + Codex into a 7-agent dev team" gets the seven-agent factory team exactly right. It understates the star count, names a team that is not one of the three main options, and lists two runtimes where OpenRig supports four.
The video comes from @buildwithneej (Neej). The repo is mvschwarz/openrig, owned by Mike Schwarz and licensed under Apache-2.0.
What is OpenRig?
The README describes it as "a local daemon + CLI + terminal UI + MCP server, built on tmux." The GitHub description reads: "Build your own network of agents from Claude Code, Codex and Pi: persistent teams with roles, shared context and owned work."
Each agent gets a seat with a job. A rig file in YAML declares the seats, groups them into pods and sets which runtime each one uses. Edges such as delegates_to define who hands work to whom.
What is the OpenRig factory team?
The video says "The factory team has seven agents, a lead, advisor, a builder, QA, a designer, and two reviewers." That matches the shipped file packages/daemon/specs/rigs/launch/factory/rig.yaml, which declares three pods:
- Orchestration:
leadon Codex, andadvisoron Claude Code - Development:
buildon Claude Code,qaon Codex, anddesignon Claude Code - Review:
r1on Claude Code andr2on Codex
The file's summary reads "A product factory with a lead, a PM advisor, build, QA, design, and two independent reviewers. Bring ideas to the advisor; the lead runs the team." The advisor is the product-management agent running under an advisor profile, which is why the README calls it a PM advisor.
What teams does OpenRig ship?
The README lists three main teams: "Three teams grow with you, and the kernel operator helps you pick one:"
starter: "a builder and a reviewer for one bounded change"workshop: "a lead, a builder, QA and a reviewer for ongoing work in one repository"factory: seven agents "for sustained product work"
The video names the three as starter, PM and factory. The PM team exists, but the README lists it separately under "Specialist teams," alongside code-review (two independent reviews) and research (an analyst and a synthesizer). The pm rig has three seats: a PM lead on Claude Code, a researcher on Codex and a builder on Claude Code. Its summary describes an 8-step feature flow that includes mockups and prototypes, so the video's description of what it does is accurate. Its place in the lineup is not.
The team the video skipped, workshop, is the one the README notes "ships with broad access."
How does the OpenRig starter team work?
The video says "a starter team has Claude build while Codex reviews every change." The starter rig file declares a build seat on Claude Code and a review seat on Codex, pinned to the gpt-6-astra model, joined by a delegates_to edge from build to review. The video matches the file.
One detail it compresses: the starter reviewer uses the QA agent spec, not the independent-reviewer spec the factory team uses for r1 and r2.
Does OpenRig only support Claude and Codex?
It supports more. The video says the repo "boots each agent into a Claude or Codex session." The README lists the runtimes as "Native Claude Code and Codex sessions, terminal nodes, and Pi and Oh My Pi via RPC runners."
The rig specification documents runtime: pi, which "launches the Pi coding agent (pi --mode rpc) through OpenRig's RPC runner," and runtime: omp for Oh My Pi, which "is separate from runtime: pi." A runtime: terminal node handles servers, log tails and build watchers, which the spec notes are "not agent runtimes." The README text visible in the video itself says "Claude Code, Codex and Pi work together as a persistent team."
What do you need to run OpenRig?
The README says: "You need Node.js 22 or 24 (22 on Apple silicon), tmux, macOS or Linux (WSL2 on Windows), and Claude Code or Codex signed in. One provider is enough to start."
The video says it "runs on the Claude and Codex accounts that you already have." That holds for the starter team, since "the kernel operator adapts the shipped starter team to whichever you have." The factory team is different. Its YAML hard-codes Codex for some seats and Claude Code for others, so running it as shipped needs both providers signed in. The kernel rig ships rig-claude-only.yaml and rig-codex-only.yaml variants, which is how the single-provider adaptation works.
Does OpenRig have a dashboard?
It has a terminal UI. The README lists its views as "Topology explorer, table and graph views, seat details, Specs, Projects, Terminals, Feed, and System." It also notes that "only the older React web UI is in maintenance mode, with best-effort support," so the maintained interface is the terminal one, not a browser dashboard.
The footage in the video is the repo's own demo. A footer in the video reads "from the repo · example data," and the README's screenshot caption says "Captured from the interactive TUI demo using fictional project data." The model names in the demo table, fable-5-1 and Opus 5.5, are real: Anthropic's models overview lists Claude Fable 5.1 and Claude Opus 5.5 as current models.
How popular is OpenRig?
The video says "This repo has over 5,000 stars." The GitHub API reported 8,168 stars, 666 forks and 147 open issues on October 24, 2026. The creator's own TikTok description says "over 8,000," so the audio contradicts the caption.
The repo went public on April 1, 2026. The npm package @openrig/cli was first published on April 6, 2026 and has 65 versions. Version 0.7.0 was released on GitHub at 14:12 UTC on October 23, 2026, about 72 minutes before the TikTok posted.
Bottom line
OpenRig's factory team is seven seats, exactly as the video says, and the starter team does pair a Claude Code builder with a Codex reviewer. The three main teams are starter, workshop and factory, with PM listed as a specialist team. OpenRig supports Claude Code, Codex, Pi and Oh My Pi. One provider is enough for the starter team, but the factory team as shipped needs both Claude Code and Codex.
Sources
- mvschwarz/openrig repository record, GitHub API - stars, forks, license and creation date
- OpenRig README - teams, runtimes, requirements and the demo caption
- factory rig specification - the seven seats and their runtimes
- starter rig specification - builder, reviewer and delegation edge
- pm rig specification - the PM specialist team
- Rig specification reference - Pi, Oh My Pi and terminal runtimes
- OpenRig 0.7.0 release - release date
- @openrig/cli on npm - publish history and license
- Models overview, Anthropic - current Claude model names
Published October 23, 2026. Based on a TikTok by @buildwithneej. Full video notes (markdown)