Crosstalk-Solutions/project-nomad

Project NOMAD is an offline-first knowledge and education server. Wikipedia, thousands of books, courses, maps, and optional local AI, all running on hardware you own with no internet required

Summary

Project NOMAD is a self-hosted, offline-first knowledge and education server written in TypeScript and licensed under Apache-2.0. It uses Docker to bundle tools like offline Wikipedia (Kiwix), Khan Academy courses (Kolibri), regional maps (ProtoMaps), a local AI assistant (Ollama plus Qdrant), and CyberChef into a single browser-accessible "Command Center." The project targets Debian-based systems (Ubuntu 26.04 LTS recommended) and is designed to run entirely on local hardware with no internet required after installation.

What it is useful for

NOMAD is useful when you need a private, local-network knowledge stack that engineers, students, or field teams can use without depending on cloud services. Concrete use cases documented in the README include: keeping an offline copy of Wikipedia, medical references, and survival guides on a lab or shop-floor machine; running Khan Academy content with progress tracking on a classroom server; querying local documents through a RAG pipeline built on Ollama and Qdrant without sending data to third parties; and using CyberChef or ProtoMaps in environments where the public internet is unavailable or restricted. The "Supply Depot" acts as a one-click app catalog (PDF tools, file browser, password manager, ebook library) and also lets you add your own custom Docker containers, so it can double as a general-purpose self-host dashboard.

How engineers can use it

The documented quick install on a Debian-based host is a single root script:

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 it finishes, you open http://localhost:8080 (or http://DEVICE_IP:8080) in a browser. The base Command Center needs about 5 GB of disk and under 1 GB of RAM; enabling the AI Assistant grows the install to roughly 25 GB and pushes memory requirements up sharply, with the README recommending 32 GB of system RAM and an NVIDIA RTX 3060 or equivalent for LLM workloads. For more control, an advanced path copies the management_compose.yaml template and runs docker compose up -d. Helper scripts in /opt/project-nomad cover start, stop, update, and uninstall operations.

A few documented caveats matter for engineers: there is no built-in authentication, and the project is explicitly not designed to be exposed to the public internet, so access control should be handled at the network level. The auto-updater can be exercised in CI with node ace auto-update:dry-run --scenarios from the admin/ directory, which simulates the full decision pipeline without triggering a real update. Major version bumps always require a manual update, and the Windows path through WSL2 is community-supported.