project-nomad

Project NOMAD is a self-contained, offline-first knowledge and education server. It is installed on Debian-based systems (Ubuntu 26.04 LTS recommended, 24.04 LTS and Debian 12 also supported) through a terminal-based script, and all tools are accessed through a browser, so no desktop environment is required. The quick install uses apt and a curl-piped install script; an advanced path provides a Docker Compose template for manual configuration. After installation the Command Center is reachable at port 8080 on localhost or the device IP. Architecture and operation: NOMAD is a management UI ("Command Center") plus API that orchestrates a collection of containerized tools and resources via Docker. It handles installation, configuration and updates of those components. A setup wizard guides first-time configuration with curated content collections, and there are built-in tools such as a Wikipedia content selector, a ZIM library manager and a content explorer. Built-in capabilities described in the README: - AI chat with knowledge base: local AI chat powered by Ollama, or OpenAI-compatible servers such as LM Studio or llama.cpp, with document upload and semantic search (RAG via Qdrant). - Information library: offline Wikipedia, medical references, ebooks and more via Kiwix. - Education platform: Khan Academy courses with progress tracking and multi-user support via Kolibri. - Offline maps: downloadable regional maps via ProtoMaps. - Data tools: encryption, encoding, hashing and analysis via CyberChef. - Notes: local markdown note-taking via FlatNotes. - System benchmark: hardware scoring with Builder Tags and a community leaderboard. - Supply Depot: a one-click app catalog (PDF tools, file browser, e-book library, password manager and more) plus the ability to run custom Docker containers. - Automatic updates: opt-in scheduled updates for core software, installed apps and offline content; minor/patch self-updates can occur in a configurable window after a cool-off period and only when pre-flight checks pass, while major versions require manual updates. Hardware: the management application itself is lightweight, with minimum specs of a 2 GHz dual-core processor, 4 GB RAM, at least 5 GB free disk space and a Debian-based OS; internet is required only during installation. Running LLMs and other AI tools is encouraged on a GPU-backed device, with optimal specs listed as a Ryzen 7 or Core i7 or better, 32 GB RAM, an NVIDIA RTX 3060 or AMD equivalent or better, and at least 250 GB free space preferably on SSD. AI models can run on a different host by pointing the AI assistant settings at an Ollama or OpenAI-compatible API URL; for remote Ollama the server must be started with OLLAMA_HOST=0.0.0.0. Privacy and connectivity: NOMAD is designed for offline usage and states it has zero built-in telemetry. Internet is needed only for initial installation and for later downloads the user chooses. Connectivity checks first try Cloudflare's 1.1.1.1 trace endpoint and fall back to endpoints the app already contacts (GitHub API and the Project NOMAD API). The test URL can be overridden in Settings under Advanced or via the INTERNET_STATUS_TEST_URL environment variable, with the environment variable taking precedence. Security posture: by design the project includes no authentication and is intended to be open on a local network; the README recommends network-level controls and advises against exposing an instance directly to the internet. Optional authentication is mentioned as a possible future feature, tracked on the public roadmap. Maintenance and development: helper scripts in /opt/project-nomad start, stop, update and uninstall the stack; the update script only refreshes Command Center containers, not installable application containers, which are updated through the UI. For contributors, an Ace command (auto-update:dry-run) exercises the entire auto-update decision pipeline without triggering an update, supporting deterministic scenario suites, live GitHub release simulation, and fully offline runs with canned release lists and a fixed clock. The project is licensed under Apache License 2.0 and offers a website, Discord community, benchmark leaderboard, FAQ and community add-ons documentation.