A curated list of content, tools, and infrastructure for keeping knowledge accessible without the internet — for disasters, shutdowns, remote regions, and the long term.
Scope. This list is about offline access to knowledge: content licensed for offline storage and redistribution, tools to read, serve, and capture it, and the ways it travels without a network.
It completes a triangle with two sibling lists, both under Other Related Lists below:
- awesome-decentralized-web — who controls the infrastructure.
- awesome-resilient-communication — how people communicate when infrastructure fails.
Out of scope:
- Pirated or unlicensed content. Everything listed is public domain, openly licensed, or free to download and redistribute offline for noncommercial use — and any entry whose license restricts commercial use or redistribution states the restriction in its description.
- Cryptocurrency- and token-dependent projects.
- AI tools and agent frameworks.
- General self-hosting software without an offline-first purpose.
Please read CONTRIBUTING.md before submitting a project.
How to read this list. Three different things get called "offline knowledge", and you usually need all three:
- Content — the knowledge itself, in bulk, licensed for offline storage and redistribution. (Wikimedia dumps, Project Gutenberg, OpenStreetMap)
- Tools — readers, servers, and capture software that make content usable off the network. (Kiwix, Calibre, ArchiveBox)
- Transport — how content moves without connectivity: hotspots, removable media, sneakernets. (Internet-in-a-Box, NNCP)
Entries marked Dormant still work but their source repository has had no activity for over 2 years — the tag states the measured date. Dead projects live in the Graveyard section at the bottom.
- Prepare Before You Need It
- Choose by Scenario
- Deployment at a Glance
- Knowledge per Gigabyte
- Content Collections
- Readers and Reference Tools
- Offline Servers and Learning Platforms
- Maps and Navigation
- Capture and Web Archiving
- Preservation, Packaging, and Verification
- Sneakernets and Offline Transfer
- Case Studies
- Guides and Communities
- Graveyard
- Other Related Lists
- Contributors
Everything on this list works only if it was downloaded, installed, and tested while the network still worked. Preparation is unglamorous and takes an afternoon; the checklist is short:
- Download content and the applications that read it in advance, and check free storage before starting a multi-gigabyte archive.
- Test the complete setup with Wi-Fi and cellular disabled: first launch, opening content, search. Confirm nothing needs an online login, license activation, DNS lookup, cloud API, or one-time network setup.
- Keep at least two verified copies on different storage devices.
- Verify checksums or signatures where the source publishes them.
- Record each archive's version and date, and choose a refresh schedule that matches how quickly the material changes and how consequential outdated information would be — maps, medical references, and software documentation all age at different rates.
- Store readers and installers alongside the content itself: an archive is useless if compatible software is not at hand for your platform.
- Keep short startup instructions, printed or as a plain text file on the same media, so someone other than you can bring the system up.
- Plan for power: charged devices, replaceable batteries or power banks, solar charging where outages may be long, and low-power servers such as a Raspberry Pi where a hub must stay running.
- If you run a community server, test it from several client devices over its own Wi-Fi, not only from the machine it runs on.
- Prefer formats that stay usable without a specific vendor or account — plain files outlive app-locked libraries.
- For medical and emergency material, note the edition date: offline references age, and none of them replace qualified professional help.
Reasonable starting combinations, drawn from the entries in the sections below — starting points, not guarantees, and the medical caveat above applies throughout.
| Scenario | Reasonable starting point |
|---|---|
| Personal offline reference library | Kiwix with ZIM archives from the Kiwix Library, Calibre for books, Zeal or DevDocs for programming documentation |
| Offline classroom | Kolibri or MoodleBox on a Raspberry Pi, with PhET simulations and OpenStax textbooks loaded in advance |
| Community knowledge hotspot | Internet-in-a-Box or RACHEL serving Wikipedia ZIM archives, offline maps, and books over local Wi-Fi |
| Medical reference collection | WikiMed plus Hesperian Health Guides, with edition dates recorded |
| Offline maps and navigation | Organic Maps, CoMaps, or OsmAnd on phones; Geofabrik extracts and Protomaps for self-hosted maps |
| Capturing websites before they disappear | ArchiveWeb.page or Browsertrix Crawler to capture, ReplayWeb.page to read the result, Zimit when the target format is a Kiwix ZIM |
| Long-term preservation | BagIt packaging with checksum manifests, Archivematica for a full workflow, guidance from digipres.org |
| Moving updates across an air gap | NNCP or plain removable media, apt-offline for Debian-based systems, checksums verified on both sides |
The entries most often deployed in bulk, compared on what matters when you provision hardware. Storage figures are approximate as of September 2026 and only grow — treat them as planning numbers and check current sizes before buying media.
| System | Archive format | Approx. storage | Update method | Incremental updates | Licensing | Platforms |
|---|---|---|---|---|---|---|
| Kiwix (ZIM archives) | ZIM | MBs per archive, up to ~119 GB for full English Wikipedia with images | Re-download the whole ZIM from the Kiwix Library | ✗ whole-file replace | Reader GPLv3; content mostly CC BY-SA | Windows, macOS, Linux, Android, iOS, Raspberry Pi |
| Wikimedia Dumps | Compressed XML/SQL | ~20 GB compressed for English Wikipedia article text | New full dumps roughly twice a month | ✗ full re-download in practice | CC BY-SA / GFDL | Raw data — any |
| Project Gutenberg | EPUB, HTML, plain text | MBs per book; on the order of 1 TB for a full mirror | rsync or HTTP mirrors | ✓ rsync transfers only changes | Public domain (US) | Standard formats — any |
| OpenStreetMap data | PBF (planet or regional extracts) | ~88 GB planet file; country extracts far smaller | Replication diffs, minutely to daily | ✓ | ODbL | Raw data — any |
| Organic Maps / CoMaps / OsmAnd | Per-region map files | Hundreds of MB per country | In-app region downloads | ◐ OsmAnd offers diff-based live updates; the others re-download regions | Apps open source; map data ODbL | Android, iOS |
| Kolibri | Kolibri content channels | GBs, varies by channel selection | Peer-to-peer sync, USB import, or online | ✓ device-to-device sync built in | Platform MIT; content licenses vary per channel | Windows, macOS, Linux, Android, Raspberry Pi |
| Internet-in-a-Box | Bundles ZIM, Kolibri channels, and OSM maps | 64 GB card to 1 TB+ drive, per content selection | Admin console re-downloads packages | ✗ whole-package replace | Open-source stack; content licenses vary | Raspberry Pi, Linux |
| RACHEL | Preloaded module bundles | Sized to device, typically hundreds of GB | Module downloads or USB from World Possible | ✗ | Mixed open licenses per module | RACHEL devices, Raspberry Pi |
✓ supported · ◐ partial or configuration-dependent · ✗ updates mean re-downloading the whole archive
Sources: Wikipedia ZIM sizes are measured from the Kiwix download server (119 GB for the 2026-08 full build with images); the planet file size and its replication diffs from Planet OSM (88 GB PBF as of 2026-09); the rsync mirroring method from Project Gutenberg's mirroring how-to. The compressed-dump and full-mirror figures are estimates, not published numbers.
Two practical consequences: on slow or metered links, prefer the systems with incremental updates (rsync mirrors, OSM diffs, Kolibri sync) and treat whole-file systems like ZIM as things you refresh occasionally by sneakernet; and check content licenses separately from software licenses — redistributing a preloaded device is a redistribution of everything on it.
What fits in a given amount of storage, using the file sizes published on the Kiwix download server in September 2026. Kiwix offers most large collections in three editions: maxi (full text with images), nopic (full text, no images), and mini (article introductions only). The text-only and mini editions carry most of the knowledge at a fraction of the size, which is what makes small budgets workable. Figures are rounded, only grow between builds, and leave no margin for the reader software itself.
| Storage budget | Suggested load | Approx. sizes |
|---|---|---|
| 1 GB (spare phone storage) | WikiMed without images fills it alone; or WikiMed mini plus Appropedia and the PhET simulations | WikiMed nopic 822 MB · WikiMed mini 155 MB · Appropedia 555 MB · PhET (English) 104 MB |
| 4 GB (small USB stick) | Simple English Wikipedia with images plus WikiMed mini and a country map; or iFixit repair manuals alone; or WikiMed with images plus Appropedia and PhET | Simple English Wikipedia maxi 3.2 GB · iFixit 3.3 GB · WikiMed maxi 2.1 GB · Kenya OSM extract 0.4 GB |
| 16 GB (microSD card) | English Wikipedia mini (the introduction of every article) plus WikiMed with images, Appropedia, and PhET | Wikipedia mini 12 GB · WikiMed maxi 2.1 GB · Appropedia 555 MB · PhET 104 MB |
| 64 GB | English Wikipedia without images plus Wiktionary and WikiMed with images; little room is left, so add iFixit or a country map only on a 128 GB card | Wikipedia nopic 49 GB · Wiktionary nopic 8.5 GB · WikiMed maxi 2.1 GB |
| 128 GB | Full English Wikipedia with images takes nearly the whole card; iFixit and PhET fit alongside it | Wikipedia maxi 119 GB · iFixit 3.3 GB · PhET 104 MB |
| 256 GB | Full English Wikipedia with images plus all of Stack Overflow, with Wiktionary, iFixit, WikiMed, and a large country map in the remaining space | Wikipedia maxi 119 GB · Stack Overflow 107 GB · Wiktionary 8.5 GB · iFixit 3.3 GB · Germany OSM extract 4.8 GB |
| 1 TB and up | Everything above plus the Project Gutenberg ZIM and either a continental map set or the whole OSM planet; this is Internet-in-a-Box or RACHEL territory | Project Gutenberg 206 GB · United States OSM extract 12 GB · OSM planet 88 GB |
Three things the numbers say: images cost more than text, since English Wikipedia grows from 49 GB to 119 GB when they are included, so text-only editions are the best knowledge per gigabyte whenever storage is tight; a medical reference is the cheapest high-value item on the list, with WikiMed's full text fitting in under 1 GB; and collections do not just grow, they jump — the Project Gutenberg ZIM went from 72 GB in its 2023-08 build to 206 GB in 2025-11, so re-check sizes before refreshing media bought for an older build.
Sources: ZIM sizes are the listed file sizes on the Kiwix download server for the newest build of each title as of 2026-09 — wikipedia_en_all_maxi_2026-08 (119 GB), wikipedia_en_all_nopic_2026-06 (49 GB), wikipedia_en_all_mini_2026-06 (12 GB), wikipedia_en_simple_all_maxi_2026-05 (3.2 GB), wikipedia_en_medicine_maxi_2026-04 (2.1 GB), wikipedia_en_medicine_nopic_2026-04 (822 MB), wikipedia_en_medicine_mini_2026-04 (155 MB), wiktionary_en_all_nopic_2026-08 (8.5 GB), stackoverflow.com_en_all_2026-07 (107 GB), gutenberg_en_all_2025-11 (206 GB), ifixit_en_all_2025-12 (3.3 GB), appropedia_en_all_maxi_2026-02 (555 MB), phet_en_all_2026-08 (104 MB). Map sizes are the current -latest.osm.pbf extracts from Geofabrik as of 2026-09 and the planet file from Planet OSM, both linked in the table above.
The knowledge itself: bulk-downloadable, with the license stated where it is not fully open.
- Kiwix Library - Catalog of ready-made ZIM archives: all of Wikipedia, Stack Exchange, Project Gutenberg, medical references, and hundreds more.
- Wikimedia Dumps - Complete database exports of Wikipedia and its sister projects (CC BY-SA), the raw material behind most offline encyclopedia tools.
Offline medical references age: record the edition date, and treat them as support for — never a replacement of — qualified professional care.
- Hesperian Health Guides - Publisher of Where There Is No Doctor and other field-medicine guides written for places without professional care; digital editions are free to download under a custom open-copyright policy whose restrictions vary by edition and can require permission for commercial use, digital or large-scale redistribution, and translation.
- WikiMed - Curated offline snapshot of Wikipedia's medical articles (CC BY-SA), maintained with WikiProject Medicine and distributed as a Kiwix ZIM and mobile app.
- OpenStax - Peer-reviewed college textbooks free to download as PDF; licenses vary by title between CC BY and CC BY-NC-SA — the noncommercial titles are not fully open.
- PhET Interactive Simulations - Math and science simulations from the University of Colorado Boulder, downloadable individually or as a complete offline website installer; licensing varies by component — the published simulations are CC BY-NC 4.0, a noncommercial restriction, while their source code is open source.
- LibriVox - Public-domain audiobooks read by volunteers, all downloadable for offline listening.
- Project Gutenberg - More than 75,000 public-domain e-books, downloadable in bulk.
- Standard Ebooks - Carefully typeset, open-licensed editions of public-domain e-books.
- Tatoeba - Openly licensed database of millions of example sentences and their translations across hundreds of languages, downloadable in full.
- Appropedia - Wiki of appropriate technology, sustainability, and development knowledge, with offline exports available through Kiwix.
- iFixit - Repair manuals for phones, computers, appliances, and vehicles, written and maintained by a large community; the guides are CC BY-NC-SA 3.0, a noncommercial restriction, and are available in full as a Kiwix ZIM.
- Software Heritage - Universal archive of software source code, preserving hundreds of millions of projects for the long term.
Software for using stored knowledge on your own device.
- Aard2 - Android dictionary and Wikipedia reader for slob-format offline archives.
- Calibre - The standard tool for e-book library management, format conversion, and reading.
- DevDocs - Combined API documentation browser for hundreds of programming tools, installable for fully offline use.
- GoldenDict-ng - Feature-rich dictionary application supporting StarDict, Babylon, and many other offline dictionary formats.
- Kiwix - Reader and server for ZIM archives on desktop, mobile, and Raspberry Pi — the standard way to use Wikipedia offline.
- KOReader - Document and e-book reader for e-ink devices, phones, and desktops.
- Zeal - Offline documentation browser for developers, using Dash-format docsets.
A box that serves knowledge to everyone nearby.
- Internet-in-a-Box - Turns a Raspberry Pi into a local knowledge hotspot serving Wikipedia, maps, books, and lessons to nearby devices.
- Kolibri - Offline-first learning platform that syncs curricula between devices without internet, by Learning Equality.
- MoodleBox - Self-contained Moodle server on a Raspberry Pi that broadcasts its own Wi-Fi network, for courses and quizzes in classrooms without connectivity.
- RACHEL - Preloaded offline education server used in schools and community centers without connectivity, by World Possible.
- CoMaps - Community-governed offline maps and navigation app built on OpenStreetMap data, forked from Organic Maps.
- Geofabrik Downloads - Daily updated OpenStreetMap extracts (ODbL) by continent, country, and region — the practical way to get map data at a downloadable size.
- OpenStreetMap - The openly licensed map of the world; its data extracts power every offline maps app on this list.
- Organic Maps - Offline maps and navigation app built on OpenStreetMap data, with detailed hiking and cycling coverage.
- OsmAnd - Highly configurable offline maps, navigation, and map-editing app built on OpenStreetMap data.
- Protomaps - Serves an entire planet of map tiles from one static file (PMTiles), making self-hosted offline maps practical.
Make your own offline copies while the network is still up.
- ArchiveBox - Self-hosted web archive that snapshots the pages you feed it into several durable formats.
- ArchiveWeb.page - Browser extension and desktop app by Webrecorder that records the pages you browse into portable WACZ archives, entirely on your own machine.
- Browsertrix Crawler - Browser-based crawler that captures whole sites, including JavaScript-heavy ones, into WACZ archives from a single container.
- HTTrack - Long-standing website copier that mirrors entire sites for offline browsing.
- Monolith - Bundles a complete web page, assets included, into a single self-contained HTML file.
- ReplayWeb.page - Viewer that opens WARC and WACZ web archives fully offline, in the browser or as a desktop app.
- SingleFile - Browser extension that saves any page as one faithful, self-contained HTML file.
- Zimit - Crawls any website into a ZIM archive readable by Kiwix.
Capture makes a copy; preservation keeps it verifiable and readable years later.
- Archivematica - Digital-preservation system by Artefactual that processes collections into standards-based, self-describing archival packages for long-term storage.
- BagIt - Informational RFC (RFC 8493, an Independent Submission) defining a file-packaging format with checksum manifests, so a collection can be verified after every copy or transfer; implementations exist for most languages.
Knowledge that travels by hand.
- apt-offline - Updates and installs Debian and Ubuntu packages on machines that never touch the network.
- NNCP - Encrypted store-and-forward file and mail exchange over removable media, sneakernets, and intermittent links.
- El Paquete Semanal - Cuba's weekly terabyte of media and software, distributed nationwide by hand on hard drives — arguably the world's largest sneakernet.
- ArchiveTeam - Volunteer collective racing to save at-risk websites; its wiki documents formats, targets, and rescue tooling.
- digipres.org - Community-maintained index of digital-preservation tools, formats, and practice.
Projects that shaped offline knowledge access but are no longer maintained. Kept for the historical record.
- LibraryBox - Portable offline file-sharing and library server, forked from PirateBox for libraries and classrooms. Discontinued (repository inactive since 2017).
- XOWA - Standalone offline Wikipedia reader with its own wiki database engine. Discontinued (repository inactive since 2022).
- awesome-decentralized-web - Peer-to-peer, federated, and local-first software: who controls the infrastructure.
- awesome-resilient-communication - Communication during shutdowns, disasters, and off-grid operation.
- awesome-web-archiving - Web archiving in depth, maintained by the IIPC.
Thanks to all contributors. Contributions are welcome — see the contributing guide above.
This work is dedicated to the public domain under CC0 1.0.