Software gets built in a room, not in a sidebar.
Software gets built in a room,
not in a sidebar.Run Claude Code, Codex and the rest in one repository, on one canvas. Watch every change land, then send the link to your team.
Claude Code
Codex
Hermes
Cursor
OpenCode
Gemini
Kimi
Your agents, one canvas- claude-1src/game.ts
- codex-1src/pong.ts
- claude-2src/App.tsx
- claude-1
- codex-1
- claude-2
- codex-2
- claude-3
- src/App.tsxApp.tsx
- src/lib/board.tsboard.ts
- src/styles/app.cssapp.css
- src/nodes/Terminal.tsxTerminal.tsx
- src/types/order.tsorder.ts
- L’Est ÉclairUn site d’un étudiant aubois répertorie les cas d’hantavirus
- PokaaÀ 19 ans, un Strasbourgeois lance un site pour suivre l’évolution du virus
- Canal 321,5 million de vues. Hanta-tracker.app, l’application signée par un étudiant de l’UTT
- Runs in the cloud
- Every provider, your own accounts
- Two agent windows at once
- One project
- Unlimited hours Start on Solo
- Everything in Solo
- Share a canvas, and anyone can watch
- Five agent windows at once
- Three projects Start on Pro
- Everything in Pro
- Agent windows you will not count
- Canvases that belong to the team, not to whoever made them
- Projects you will not count
- Features before anybody else
Nothing to install · your agent accounts, your repository
Follow us[](https://x.com/murmell%5F)[](https://www.linkedin.com/company/137124077/)
Launch offerFree now\+ $60 of Claude credit\*75spots left
MurmellGames night_A_ M JShare
╭─ Claude Code ─────────────────╮ │ Build Snake for the arcade │ ╰───────────────────────────────╯ ✦ Creating src/game.ts ✓ game loop ● tuning keyboard input ▌
⏵⏵ autoshift+tab to cycle
kimi ~/arcade · main
▌ build the shared arcade UI
claim src/App.tsx granted edit src/App.tsx +12 −3 ▌
kimi · mainboard · 0 collisions
› Build Pong and deploy the preview
• Read src/App.tsx • Edited src/pong.ts ✓ Preview running on :5173 ▌
workspace-write100% context left
arcade.locallive
Snake
Pong03 07
AdaMalikJun
−72%+
Your agents, your team, and the repository they share, in one place.
The whole thing, in one take.
Play the demo2:51
One canvas, two browsers, and the agents doing the work. Recorded in one pass, nothing staged.
You know these three evenings.
claude-1codex-1src/App.tsxoverwritten
Two agents. One file. No rules.
claude-1 · tuning keyboard input✓ game loopclaude-1 · SIGKILL
The agent dies with your laptop.
<<<<<<<>>>>>>>
Four clones, and a Friday of merges.
You say it. Murmell starts them.
Murmell has an agent of its own. Tell it what you want built and it opens the terminals, hands each one its share of the repository, and gets out of the way.
hold to talk
spawn three agents on the arcade and wire the preview
An agent asks before it writes.
Every agent claims the file it is about to touch. One holder per path, and everybody else waits. Nothing is merged at the end because nothing forked at the start.
reservations
the board
the boardclaude-1codex-1kimi-1opencode-1src/types/order.tsorder.tsclaude-1codex-1kimi-1opencode-1
One holder per path. The board refuses everybody else until it is released.
claude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.ts
claude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.tsclaude-1_writes_src/App.tsxcodex-1_refused_src/App.tsxclaude-2_writes_src/lib/board.tscodex-2_writes_src/styles/app.cssclaude-3_refused_src/App.tsxcodex-1_writes_src/nodes/Terminal.tsxclaude-1_writes_src/types/order.tscodex-2_refused_src/lib/board.ts
\> the problem
One agent is a tool.
Five is a traffic jam.So everyone runs one agent, alone, in their own copy. The moment the work gets big enough to be worth splitting is the moment the tooling stops helping.
Murmell puts them in one directory and makes them ask before they write.
The agents are already yours.
Murmell launches the command line tools you already pay for, on its own machines. Nothing is resold and no key is stored.
Claude Code
Codex
Kimi
OpenCode
& more
Claude Code
Codex
Kimi
OpenCode
& more
Claude Code
Codex
Kimi
OpenCode
& more
FoxLeaapp.murmell.com/c/x7k2AAnaeditingOOtterviewingSSamviewingHHeroneditingBBrunoviewing
Send the link. They are in the room.
Watching costs nothing and needs no account. Cursors, names, and the terminal output as it lands, for as many people as you send it to.
Press
Follow us[](https://x.com/murmell%5F)[](https://www.linkedin.com/company/137124077/)
Hours are never the limit.
A machine costs the same whether you use it for an hour or all month, so we do not meter time and we do not count your windows. Leave an agent running.
Free now
One week on us with Pro or Builder. Nothing is charged until it is up.
Billed
MonthlyYearly
Solo
$39/month
One project, one agent account, on our machines.
Most chosen
Pro
$69/month
1 week free
Send the link. This is where the room opens.
Builder
$149/month
1 week free
For teams keeping several things warm at once.
One week free on any plan. Watching a canvas that is running needs no account; coming back to one that is off needs Pro.
\> faq
Straight answers.
What is Murmell?
Murmell is a browser canvas where several AI coding agents (Claude Code, Codex, Kimi, OpenCode) run on a cloud machine and share one project directory. Each agent opens in its own window with a real terminal, and everyone on the link sees the same canvas as it happens.
Can several people really use one canvas at the same time?
Yes. Everyone on the link sees the same room: who joined, which window they are watching, where their cursor is. Each person drives their own agents, and nothing has to be replayed or refreshed to stay in step.
Where does my code live?
On a machine Murmell runs, in one directory, and in a private GitHub repository created with the project. The workspace is disposable and the repository is not: the work is pushed before a machine is ever torn down.
How much does Murmell cost?
Murmell is a paid product: Solo $39, Pro $69, Builder $149 per month, billed monthly, or ten months' price for a year. Every plan starts with 7 days free; Pro and Builder also come with $60 of Opus credit. Read links, watching a canvas somebody shared, are free and uncapped, and cost the viewer nothing and no account. Murmell charges for the machines and the room, never for model tokens: you connect the agent subscriptions and API keys you already pay for.
Do you resell tokens?
No. You connect the agent subscriptions and API keys you already pay for. Murmell charges for the machine and the collaboration, never for the model.
What stops two agents destroying each other?
An agent claims the paths it is about to touch before it writes. A watcher attributes every write back to whoever made it, and a colliding write is snapshotted before it lands, so nothing is lost. The rule is enforced by the machine rather than suggested to the model.
Is each project isolated?
Yes. Every canvas runs on its own machine, separate from every other project Murmell hosts. The machine that executes agent code holds no database and no decryption key.
Which coding agents does Murmell support?
Claude Code, Codex, Kimi and OpenCode today, each in its own terminal on the canvas, signed in with your own account or API key. Adding an agent is adding a tile: nothing about the canvas changes.
Can I self-host it?
No. Murmell runs the machines for you. That is the product. Your code still ends up in a private repository you own.
\> enter the room
Build it together.
Open a canvas, bring your agents, send the link.