Features

Where your agents work.

They read the ticket themselves and build what the design shows. They check the result in the browser, open the pull request and leave the app running for you. All of it on your own machine, in a workbench that fits in one file. The most important parts are here, and the docs have the full reference.

From an idea to a running app

Starts before there is a ticket

★ Not in Codespaces or a bare devcontainer.

No ticket yet? Describe your idea and answer the questions. Discovery writes the brief, the journeys and the backlog into Confluence and Jira for you.

Docs →

Reads the ticket itself

★ Not in Codespaces or a bare devcontainer.

Say "implement ACME-1234" and it opens the issue, its description and its comments itself: Jira and Confluence, GitHub and GitLab issues, Notion, Linear.

Docs →

Builds what the design says

★ Not in Codespaces or a bare devcontainer.

When the screen exists in Figma or Claude Design, it reads the frame: the layout, the components from your design system, the tokens behind the colours.

Docs →

Knows the stack before it starts

★ Not in Codespaces or a bare devcontainer.

It starts briefed on this workbench: which languages and services are here, and how to reach the database. So it connects the app instead of inventing a URL.

Docs →

Plans it, builds it, reviews it

★ Not in Codespaces or a bare devcontainer.

Three agents pass the work along. One writes the plan, one builds exactly that, and one reviews the result before it ever reaches you.

Docs →

Checks the app in the browser

★ Not in Codespaces or a bare devcontainer.

A browser in the workbench lets it open the app, click through it and read the console and the network. It does not ask you if the page works.

Docs →

Works with git the way you do

GitHub, GitLab and Bitbucket are connected, so it does the whole round: clone, branch, commit under the right name, push, open the pull request.

Docs →

Keeps running, phone included

★ Not in Codespaces or a bare devcontainer.

The app keeps running after the session ends, at a name you can type. One command puts it on your network over HTTPS, so you open it on your phone.

Docs →
And it is your environment

You work in it, not just the agent

Attach VS Code, VS Codium, JetBrains or the Claude Code app over SSH, with nothing to set up. The workbench your agent works in is the one you open.

Docs →

Every project in its own box

Three client projects side by side, each with its own Node version, database and logins. Your machine keeps none of it, so there is nothing to uninstall.

Docs →

One file you can read

★ Not in Codespaces or a bare devcontainer.

Everything the agent got is written down in one readable file: languages, services, repos, the servers it reaches. Nothing hides in some tool setting.

Docs →
Yours, not rented

Local, no cloud, no account

★ Not in Codespaces or a bare devcontainer.

It all runs on your machine. No rental, no hourly bill, no sign-up, and nothing leaves the device unless you send it somewhere on purpose.

Docs →

Pick it up again months later

One command and the old project runs again: same toolchain, same services, still signed in. You do not spend an afternoon getting it to work first.

Docs →

Standard parts, wired together

Dev containers, Docker Compose, MCP, SSH: we invented none of it. Monoceros connects the standard parts so that they work as one workbench.

Docs →
See the full feature reference in the docs →