multi-agent-shogun, a hierarchical multi-agent orchestration system themed around samurai roles, has been released as open source on GitHub, drawing attention as a way to run up to 10 AI coding agents in parallel on top of tmux.
Open Source · multi-agent-shogun
A Feudal Chain of Command for AI Coding Agents
A new GitHub tool runs up to eight coding CLI agents in parallel under a Shogun-Karo-Ashigaru hierarchy — coordinating via tmux and YAML files for "zero coordination overhead," with no API calls.
8
agents running in parallel (up to 10 in examples)
7
supported coding CLIs under one command
$0
API tokens — communication via YAML + tmux
Cost to run 8 Opus-class agents for one hour
Pay-per-API vs. a fixed-fee CLI subscription — drawn to scale
$100+/hr
API coordination
~$200/mo
flat CLI subscription
A fixed monthly fee lets users run agents without counting hours.
The Chain of Command
Shogun — general
Receives user instructions and delegates
↓
Karo — retainer (+ Gunshi, strategist)
Breaks down and assigns tasks; plans strategy
↓
Ashigaru ×N — foot soldiers
Parallel execution workers; skills discovered bottom-up
Seven coding CLIs, one command structure
Claude Code
OpenAI Codex
GitHub Copilot
Kimi Code
OpenCode
Cursor
Antigravity
Why it appeals
Specialize roles instead of overloading one agent
Sidesteps context-window bloat & hard task splits
Memory MCP keeps state across sessions
Real-time dashboard of agent activity
Watch-outs
Learning curve for tmux management
Behavior differs across CLIs
Claude Code needs OAuth + permissions bypass
Fresh release — large-scale usage yet to come
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…