Skip to content

Wifi drops. Laptop closes. Your agents keep working.

A persistent development machine for you and your coding agents: always on, isolated from your laptop, one per project. Start a task on the train and pick it up from the couch.

Free and open source. Bring your own server.

The daemon on a train seat, typing on a laptop by the window while the landscape rolls past; a dashed lime line links the laptop to the workspace far away.

Nothing stops

Tasks, servers and agents keep running while your laptop sleeps.

Your laptop stays yours

Agents never touch your personal files, sessions or credentials.

Same workspace tomorrow

Dependencies, caches and data are still there when you return.

Lose the signal. Keep the build.

Long tasks, dev servers, workers and databases run on the workspace, not on your laptop. Whatever happens to your connection or your battery, the work continues, and you can reach it from any device.

Four beats: a laptop and phone on a train tray with no signal, the laptop closed, the workspace box still running with a terminal and a database, and a phone on a sofa arm showing a check mark.

Browser

Open a terminal and check progress from any machine.

IDE + SSH

Work normally from VS Code, Cursor or your terminal.

Phone

Check agents, restart processes or continue a session while away.

The daemon standing with crossed arms in front of a desk with a laptop, a photo, a piggy bank and keys; beyond a dashed line, a wireframe cube holds only a folder and a terminal.

Keep agents away from your laptop.

Coding agents are powerful because they can execute commands, read files and use credentials. Give them their own machine instead of yours.

Isolated environment

No access to your personal files, browser sessions or local credentials.

Disposable when needed

If an agent breaks the environment, recreate the workspace instead of your laptop.

Controlled access

Only put the repositories, credentials and services into the workspace that the agent actually needs.

Run real software,
not just agents.

The agent works inside a complete development machine, with everything a real project needs.

Install your stack

Docker, Postgres, Redis, npm, Composer, browsers and anything else you need.

Run the application

Frontend, backend, workers, databases and dev servers.

Keep the state

Dependencies, caches and development data remain between sessions.

A glass workspace box holding a terminal, a browser preview, a database, container crates and a queue, all wired together and running, with nobody inside.
Docker
Postgres
Redis
npm
Composer
Browser

One workspace per project.

Client A never sees client B. Every project gets its own workspace with its own repositories, credentials and services. An agent only reaches the workspace it runs in.

Separate credentials

API keys, tokens and .env files live in exactly one workspace.

Separate data

Databases, uploads and caches never mix between projects.

Hand over cleanly

Finish a project, download what you need, then delete its workspace.

Three wireframe cubes, each holding one project's folder with its own key, database or server, separated by dashed lines; the daemon in front holds a ring with three keys.

Switch agents. Keep the machine.

Claude Code, Codex, Cursor, OpenCode and whatever comes next, on your own accounts and subscriptions. Try a new agent without rebuilding your environment.

The workspace belongs to you, not to an agent vendor.

Claude Code
Codex CLI
Cursor CLI
OpenCode
Antigravity CLI
Grok CLI
Pi
>_ Any shell
The daemon holding the glowing end of a cable, about to plug it in.

Stop making your laptop the server.

Free and open source. Bring your own server.

An overheating laptop on the left, a calm glowing workspace box on the right, joined by a lime arrow.