Marimo | Support by cloudimg

This is a repackaged open source software product wherein additional charges apply for cloudimg support services.

Marimo is an open source reactive notebook for Python that eliminates the hidden state and stale outputs common in traditional .ipynb-based notebooks. Unlike JSON-based notebook formats, marimo notebooks are stored as pure Python files that produce readable diffs in git and run directly as scripts. Cells execute reactively - change one value and only the dependent cells re-run - so your outputs are always consistent with your code. This AMI delivers marimo fully installed and configured as a system service, giving you a production-grade reactive notebook environment running within minutes of launch.

• Pure Python storage - No JSON blobs; your notebooks produce clean, reviewable pull requests without special tooling
• Reactive execution - No need to manually "Run All" or worry about cell execution order; dependencies are tracked automatically
• Dedicated data disk - Your notebook workspace lives on a separate, independently resizable EBS volume rather than the root volume, so data survives instance replacement
• Built-in interactivity - Sliders, tables, plots, and SQL integration are native, not bolt-on extensions
• One-click web apps - Serve any notebook as a read-only interactive application for your team

• Marimo installed into a dedicated virtualenv and run by an unprivileged service account
• Notebook workspace on a dedicated data disk (independently resizable EBS volume)
• A systemd service that starts the notebook server on boot and restarts it on failure
• An nginx reverse proxy publishing the notebook editor on port 80 with WebSocket support for live reactive updates

• Amazon S3 - Store and retrieve notebooks and datasets from S3 buckets using Python libraries like boto3 within your reactive cells
• Amazon CloudWatch - Monitor instance health and notebook server uptime using standard CloudWatch metrics and alarms
• EBS Volumes - The dedicated data disk leverages EBS for resizable, snapshot-capable notebook storage
• Compatible with popular data libraries - Use Pandas, Polars, Plotly, and other Python packages directly in reactive cells

On first boot, a one-shot service generates a fresh access token unique to your instance, writes it into the marimo service configuration, and stores it in a root-only file. The notebook server binds to loopback only and is never exposed without the token. No shared or default credentials ship in the image.

The notebook editor is served on port 80 through nginx. Sign in with the generated access token to open the seeded sample reactive notebook, create your own notebooks, run cells, and explore the file browser. Any notebook can be served as an interactive app for your users.

• Data engineering teams - Build and serve daily ETL dashboards as internal web apps, with reactive cells ensuring reports always reflect the latest query results
• Education and training - Instructors can deploy individual instances for students, each with isolated workspaces on dedicated data disks, demonstrating Python with live reactive examples
• Reproducible data analysis - Analysts produce reports where outputs are guaranteed consistent with code, eliminating "works on my machine" issues common with execution-order-dependent notebooks
• Interactive data apps - Prototype and deploy internal tools with sliders, tables, and plots without leaving the notebook environment
• Exploratory analysis - Version-controllable notebooks that diff cleanly make collaborative data science practical

24/7 technical support by email and live chat. Our engineers help with deployment, configuration, the reactive execution model, UI elements, SQL integration, serving notebooks as apps, package management, TLS setup, and runtime tuning.