OpenRig

Orchestrate and coordinate teams of autonomous AI agents across machines

PlayableOpen source

Description

OpenRig is an agentic framework designed for coordinating teams of AI agents using diverse model pods. It features a Terminal UI (TUI) that allows users to monitor agent organization, track workflow progress, and manage coordination across multiple machines in one centralized workspace.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: A sophisticated multi-agent framework with a dedicated TUI for visualizing and managing complex agent topologies.
Attributed to: @esotericlabs
Built with: Codex GPT-6 Astra
Links: openrig.dev
View tweet
View GitHub repo
Tags: #agents #tui #orchestration #multi-agent

Discovered: X
First posted: X · Oct 5, 2026
Listed: Added just now
Build: Strong
Models: Moderate

Prompt

Set up OpenRig. Build a lean team with three seats: a planner, a worker, and a checker.

Step 1. Prepare the machine.

→ Run `node -v` and `tmux -V`.
→ OpenRig needs Node 22 or 24, and tmux.
→ Read the section "what OpenRig changes on your machine" in the OpenRig repository.
→ Tell me the files that OpenRig writes.
→ Back up your Codex and Claude Code configuration.

Step 2. Install OpenRig. Create the team file.

→ Run `npm install -g @openrig/cli`.
→ Run `rig setup --dry-run`. Show me the plan.
→ Apply the plan with `rig setup`.

→ Create the file `~/.openrig/specs/lean-team/rig.yaml`.
→ Use the built-in agents from `rig specs ls --json`.
→ Add three seats:

→ `lead` (orch pod): claude-code, effort high. This seat plans the work.
→ `builder` (dev pod): codex, effort low, cheapest fast model. This seat does the work.
→ `check` (dev pod): codex, effort high. This seat checks the work.

Connect the seats:

→ The lead sends work to the builder.
→ The lead sends work to the check.
→ The builder sends work to the check.
→ The check watches the builder.

Put the edges of one pod in the pod block. Put the cross-pod edges in the rig block.

Step 3. Start the team. Give it one job.

→ Run `rig specs add ~/.openrig/specs/lean-team`.
→ Run `rig specs preview lean-team --kind rig`.
→ Correct each error that the command shows.

→ Go to your repository. Run `rig up lean-team`.
→ Run `rig tui --shared` to open the dashboard.

→ Run `rig ps --nodes --rig lean-team`. Answer each permission question.

Send one small change to `orch-lead@lean-team`.

Report the seat table, the task number, the diff, and the test steps.

Source

Explore

Agents / Skills

OpenRig

Orchestrate and coordinate teams of autonomous AI agents across machines

byEsoteric Labs

PlayableOpen source

Description

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: A sophisticated multi-agent framework with a dedicated TUI for visualizing and managing complex agent topologies.
Attributed to: @esotericlabs
Built with: Codex GPT-6 Astra
Links: openrig.dev
View tweet
View GitHub repo
Tags: #agents #tui #orchestration #multi-agent

Discovered: X
First posted: X · Oct 5, 2026
Listed: Added just now
Build: Strong
Models: Moderate

Prompt

Set up OpenRig. Build a lean team with three seats: a planner, a worker, and a checker.

Step 1. Prepare the machine.

→ Run `node -v` and `tmux -V`.
→ OpenRig needs Node 22 or 24, and tmux.
→ Read the section "what OpenRig changes on your machine" in the OpenRig repository.
→ Tell me the files that OpenRig writes.
→ Back up your Codex and Claude Code configuration.

Step 2. Install OpenRig. Create the team file.

→ Run `npm install -g @openrig/cli`.
→ Run `rig setup --dry-run`. Show me the plan.
→ Apply the plan with `rig setup`.

→ Create the file `~/.openrig/specs/lean-team/rig.yaml`.
→ Use the built-in agents from `rig specs ls --json`.
→ Add three seats:

→ `lead` (orch pod): claude-code, effort high. This seat plans the work.
→ `builder` (dev pod): codex, effort low, cheapest fast model. This seat does the work.
→ `check` (dev pod): codex, effort high. This seat checks the work.

Connect the seats:

→ The lead sends work to the builder.
→ The lead sends work to the check.
→ The builder sends work to the check.
→ The check watches the builder.

Put the edges of one pod in the pod block. Put the cross-pod edges in the rig block.

Step 3. Start the team. Give it one job.

→ Run `rig specs add ~/.openrig/specs/lean-team`.
→ Run `rig specs preview lean-team --kind rig`.
→ Correct each error that the command shows.

→ Go to your repository. Run `rig up lean-team`.
→ Run `rig tui --shared` to open the dashboard.

→ Run `rig ps --nodes --rig lean-team`. Answer each permission question.

Send one small change to `orch-lead@lean-team`.

Report the seat table, the task number, the diff, and the test steps.

Source

OpenRig

Description

How it grew

OpenRig

Description

How it grew