Marimo Hub — Project and Environment Management for Marimo

[Samwise333]

Marimo Hub — Project and Environment Management for Marimo

Hello Marimo community! 👋

I'd like to share a solution to a problem that many of us have probably encountered: Marimo doesn't have built-in support for quickly switching between virtual environments.

🤔 The Problem

When working with multiple Marimo projects, each requires its own dependencies and Python version. In the standard scenario, you have to:

• Manually create and switch between virtual environments
• Run marimo edit with different environments
• Remember which environment belongs to which project
• Manage passwords and settings for each project separately

This works, but it's not very convenient, especially when the number of projects grows.

✨ The Solution: Marimo Hub

I've created a prototype Marimo Hub — a web application for managing Marimo projects, virtual environments, and users with a convenient interface.

Key Features

1. Project Management

• Create and organize your Marimo projects in one place
• Each project has its own directory and can contain multiple environments
• Support for descriptions and metadata for projects

2. Flexible Environment Management

• Support for different package managers: UV and Conda or others through templates
• Choose Python version for each environment
• Automatic Marimo installation in every new environment
• Environment templates for quick creation of standard configurations
• Easy switching between environments of the same project or different projects

3. User Management and Security

• Authentication system with JWT tokens
• Each user has their own projects
• Automatic generation of secure passwords for Marimo
• Roles and access rights (expansion planned)

4. Convenient Web Interface

• Modern UI built with Vue(TS) + TailwindCSS + Vite
• Real-time notifications via WebSocket
• Marimo process management through the interface
• View statuses of running processes
• Quick access to projects and their environments

5. Automation and Monitoring

• Automatic start and stop of Marimo processes
• Process status tracking
• Cleanup of stale processes
• Logging and diagnostics

🏗️ Tech Stack

Backend:

• FastAPI (Python 3.14+)
• SQLAlchemy for database operations
• UV for Python dependency management
• WebSocket for real-time communication
• JWT for authentication

Frontend:

• Vue 3 + TypeScript
• Vite for fast development
• Modern UI

Infrastructure:

• Docker support for deployment
• Taskfile for command automation
• Structured logging

💡 How It Solves the Problem

Before Marimo Hub:

# Manual switching between environments
source project1/.venv/bin/activate
marimo edit --port 8000
# ... work ...
# Need another environment? Stop, deactivate, activate new one
deactivate
source project2/.venv/bin/activate
marimo edit --port 8001

With Marimo Hub:

1. Open the web interface
2. Select a project
3. Choose an environment (or create a new one)
4. Click "Start Marimo"
5. Done! The process is automatically started with the correct environment

All switching is just a few clicks, no terminal needed!

🎯 Use Cases

Use Case 1: Multiple Projects

You're working on:

• A data analysis project (pandas, numpy, matplotlib)
• A machine learning project (scikit-learn, torch)
• A web scraping project (requests, beautifulsoup4)

Each project has its own environment with required dependencies. With Marimo Hub, you easily switch between them through the interface.

Use Case 2: Different Python Versions

One project needs Python 3.11, another needs 3.12. You create environments with the required versions, and the system automatically uses the correct interpreter.

Use Case 3: Team Collaboration

Each team member has their own account and projects. Projects can be easily transferred between users (expanded functionality planned).

🔮 Future Plans

• Collaborative project work (sharing)
• Git integration for versioning templates and projects
• Project templates from a gallery
• Export/import of environment configurations
• Support for additional package managers
• API for automation and CI/CD
• Deployment via helm and docker

💬 Discussion

I'd love to hear your thoughts:

• Is such a tool needed? :)
• Have you encountered the problem of switching between environments?
• What would you add to such a solution?
• Are there other pain points when working with Marimo that could be solved?

---

P.S. If anyone is interested, I can provide a more detailed architecture overview or show how it works under the hood.

[stephenlf]

I love it! I have set up a shared Marimo box for my team at work. I tried to use the JupyterHub Marimo Proxy to configure an isolated, multi-user setup, but got buried in integration hell and ultimately couldn't make it work. I ended up giving each user a separate Linux user on the machine and spun up a separate Marimo process for each user. It works, but it's not great. Something like this would have been much easier to manage.

Here are some pain points from my solution that I never resolved. Does your solution address any of these?

• Poor security. I am only using Marimo's built-in token authentication, which a tricky user could snoop.
• Difficulty sharing notebooks. We share notebooks through a shared directory symlinked into each user's project root. A recent Marimo security update broke that system. (See Sharing notebooks on the same machine? I.e. workaround for `Access denied: File is outside the allowed directory.` #7655)
• Server management is hard. Users don't update their servers. They don't spin down their notebooks. They eat through RAM.