Assign the right work
MUON turns a goal into clear jobs with owners, limits, and checkpoints before anyone starts coding.
Multi-agent software teams, under control
Make your coding agents work as one.
You already have AI coding tools. The hard part is making them work together without becoming your full-time job. Give MUON one mission. It plans the work, keeps the crew aligned, and brings you back when a real decision is needed.
Works with the tools you already useShared context across the crewYou approve what ships
Free download. Bring the AI tools you already pay for.
See how it worksOLLAMA
Optional local recallNot a coding workerCLAUDE CODE
Implementation / reviewCHILD 01CHILD 02
CENTRAL ORCHESTRATION
MUON
MUONPlans · assigns · coordinates · governs
UNIFIED CREW MEMORY
CURSOR
Managed read-onlyScout / review laneCODEX
Implementation / reviewCHILD 03CHILD 04
HUMAN AUTHORITYApprove fan-out · merge · shipOne mission record
The coordination gap
More agents can mean more code. They also mean more briefings, handoffs, collisions, and review work for a person to manage.
MUON handles the coordination so your engineers can lead instead of babysit.
Keep the crew aligned
Agents get the context they need, see what teammates already tried, and avoid stepping on the same work.
Review one result
You get one place to inspect the change, the checks, and the history before you decide it can land.
One mission, one coordinated crew
How MUON runs the crew.
MUON turns a request into assigned work, keeps agents aware of the mission and each other, and returns one reviewable result to your team.
Plan
Turn a goal into clear jobs
MUON records what you want done, breaks it into owned steps, and suggests who should do each part.
Assign
Put the right tool on the right job
Claude Code and Codex can build. Cursor and OpenCode help with exploration and review. Every job has limits and a clear owner.
Coordinate
Keep everyone on the same page
Shared memory, handoffs, and live updates move with the work so you are not copy-pasting between chat windows.
Review
Decide with the full picture
Your team reviews the change, the checks, and the trail before anything merges into the main project.
Shared context without shared confusion
Stop copying context between agent windows.
Each agent keeps memory for its own job. MUON also shares what the crew already knows so people are not copy-pasting context between windows.
Personal agent memoryShared crew memoryYou confirm lasting knowledgeStays on your machine
Explore shared memoryUse MUON where you work
One mission, wherever your team works.
Move from the Desktop app to the terminal or a connected Claude or Codex chat without splitting jobs, approvals, or history.
Desktop
Full workspaceThe visual home for missions, the crew, memory, approvals, and review in one place.
Terminal
Same controlRun the same missions from the command line or a compact terminal UI when you prefer the keyboard.
Your AI chat
Optional leadStart Claude Code or Codex yourself, connect it to MUON, and let it lead the crew while you still approve what ships.
A role for every tool
MUON runs the crew. Each tool has a clear job.
Claude Code and Codex can build and lead. Cursor and OpenCode help explore and review. MUON keeps those differences honest instead of pretending every tool is the same.
Claude CodeCodexCursorOpenCode
Build and lead
Claude Code
ReadyCan plan, implement, review, and lead a MUON crew from its own chat when you connect it.
Build and lead
Codex
ReadySame strength as Claude Code for building and leading. Connect it from your own Codex session when you want that workflow.
Explore and review
Cursor
ReadyDeep code exploration and sharp review passes, pointed exactly where the crew needs a second opinion.
Explore
OpenCode
ReadyFast reconnaissance that maps the terrain and gathers context before the builders move.
Trust
More agent output without giving up control.
Your work stays on your machine
MUON runs locally. The core loop does not depend on sending your engineering process through someone else's cloud.
Permission stays narrow
Approvals cover a specific action for a specific job. Shipping and other high-impact steps always come back to a person.
Delivery stays reviewable
What changed, what was checked, and who decided stay attached to the mission so you are not reconstructing the story later.
