Does Project N.O.M.A.D. require an internet connection to run?

No. Internet is only needed during initial installation to download dependencies and later if you choose to download more content. Otherwise it runs fully offline with no built-in telemetry.

Is there user authentication or access control?

No. By design, N.O.M.A.D. ships without built-in authentication. Access control is expected to be handled at the network level (firewall/port rules), and the project advises against exposing it directly to the internet.

What hardware do I need?

The Command Center itself runs on modest hardware (2 GHz dual-core, 4 GB RAM, 5 GB disk), but running the AI assistant well requires a GPU-backed machine, ideally 32 GB RAM and an RTX 3060-class GPU or better with 250 GB storage.

Can I run the AI model on a separate machine?

Yes. You can point the AI assistant settings to a remote Ollama or OpenAI-compatible API server (e.g. LM Studio). If using Ollama remotely, start it with `OLLAMA_HOST=0.0.0.0`.

What operating systems are supported?

Any Debian-based OS, with Ubuntu recommended for the native install script. Windows users can use WSL2 with Docker or Docker Desktop, following the community-supported guide.

What license is the project under?

Project N.O.M.A.D. is licensed under Apache License 2.0.

Nomad: Self-hosted offline knowledge

Project N.O.M.A.D. (Node for Offline Media, Archives, and Data) is a self-hosted management platform that orchestrates a set of containerized offline tools: a local AI chat assistant, offline Wikipedia and reference libraries, an education platform, offline maps, data analysis utilities, and note-taking. It's built for people who want a browser-accessible knowledge and education server that keeps working with no internet connection, whether that's for disaster preparedness, remote locations, homeschooling, or general privacy-focused self-hosting.

The project ships a "Command Center" (a TypeScript-based management UI and API) that handles installing, configuring, and updating everything under the hood via Docker. You interact with it through a browser, so a full desktop environment isn't required if you'd rather run it headless as a server.

Key features

AI Chat with Knowledge Base: Local AI chat via Ollama (or any OpenAI-compatible server like LM Studio), with document upload and semantic search using Qdrant for RAG.
Information Library: Offline Wikipedia, medical references, survival guides, and ebooks served through Kiwix.
Education Platform: Khan Academy courses with progress tracking and multi-user support via Kolibri.
Offline Maps: Downloadable regional maps for offline viewing and search via ProtoMaps.
Data Tools: Encryption, encoding, hashing, and analysis through a built-in CyberChef instance.
Notes: Local markdown note-taking via FlatNotes.
System Benchmark: Hardware scoring with Builder Tags and a public community leaderboard.
Supply Depot: A one-click app catalog (PDF tools, file browser, password manager, and more) plus support for running your own custom Docker containers.
Automatic Updates: Opt-in, scheduled updates for the core software, installed apps, and offline content.
Easy Setup Wizard: Guided first-time configuration with curated content collections.

Ideal use cases

N.O.M.A.D. fits scenarios where you want a single device holding a large offline knowledge base and tools reachable from any browser on a local network: emergency preparedness kits, off-grid cabins, boats or RVs, homeschool setups, or a home lab that doubles as a family reference library with an AI assistant. It also works well for hobbyists who want to benchmark hardware against a community leaderboard or run a custom app catalog alongside offline content.

It's a poor fit if you need multi-user authentication or role-based access out of the box: the project ships with no built-in auth, by design, and relies on network-level controls instead. It's also not meant to be exposed directly to the internet. If your priority is a lightweight, low-power device, note that the AI features specifically want a GPU-backed machine with real RAM and storage; the barebones Command Center is light, but the useful parts (LLMs, large offline libraries) are not.

Installation

N.O.M.A.D. installs on any Debian-based OS (Ubuntu recommended) via a terminal-based script. Root/sudo access is required.

Quick install:

sudo apt-get update && \
sudo apt-get install -y curl && \
curl -fsSL https://raw.githubusercontent.com/Crosstalk-Solutions/project-nomad/refs/heads/main/install/install_nomad.sh \
  -o install_nomad.sh && \
sudo bash install_nomad.sh

After install, open http://localhost:8080 (or http://DEVICE_IP:8080) in a browser to start the setup wizard.

For more control, use the Docker Compose template directly: copy the provided management_compose.yaml into a docker-compose.yml, customize placeholders, then run:

docker compose up -d

This path is recommended only for users comfortable with Docker and manual configuration.

Once installed, helper scripts live in /opt/project-nomad:

sudo bash /opt/project-nomad/start_nomad.sh    # start all containers
sudo bash /opt/project-nomad/stop_nomad.sh     # stop all containers
sudo bash /opt/project-nomad/update_nomad.sh   # update Command Center containers

To uninstall cleanly:

curl -fsSL https://raw.githubusercontent.com/Crosstalk-Solutions/project-nomad/refs/heads/main/install/uninstall_nomad.sh -o uninstall_nomad.sh && sudo bash uninstall_nomad.sh

Minimum hardware is a 2 GHz dual-core CPU, 4 GB RAM, and 5 GB free disk. To run the AI tools well, the project recommends something closer to a Ryzen 7/Core i7, 32 GB RAM, an RTX 3060-class GPU, and 250 GB of SSD storage. Internet access is only needed during install and when downloading additional content later.