Analytics cookies only (Google Analytics, PostHog). No ad networks, no cross-site tracking. See our privacy policy.
Self-host ArchiveBox
updated Nov 2026deployed by usprices checked · Jul 2026Install verified on Ubuntu 26.04 · Nov 2026 · how we test
We earn commissions when you shop through the links below. Full disclosure →
A self-hosted web archive. You feed it URLs, browser history, bookmarks exports or RSS feeds, and it saves each page in several formats (HTML, a single-file copy, PDF, screenshot, WARC, media through yt-dlp) so a link still works after the original disappears. It has a web interface with search and tags, a CLI and a browser extension. It saves whole pages rather than reading-mode text, so it suits keeping evidence and references more than a daily reading queue, and the archive grows large.
Pick ArchiveBox when you want to keep a link after the page it points to changes or vanishes, not just read it later. You give it URLs, browser history or a bookmarks export and it saves each page in several formats: HTML, a single-file copy, PDF, screenshot, WARC, and audio or video through yt-dlp. It has a web interface with search, a CLI and a browser extension. It is MIT-licensed, and our snippet runs the `archivebox/archivebox` image pinned to 0.9.77 (released 2026-10-27), adapted from the project's own Compose file, with an admin account created from environment variables. We did not find an upstream RAM minimum, so our 2048 MB floor is an estimate based on the Chrome instance each snapshot runs in; the Compose file asks for a 1 GB shared-memory pool for it. In the catalog, Wallabag and Readeck store a cleaned, readable copy of an article for a reading queue; ArchiveBox stores the whole page for preservation and is slower and much larger. Choose it for references, research and evidence, and one of the other two for a daily reading list.
What you need
Any VPS with at least 2048 MB of RAM
A domain you control — most self-hosted setups need HTTPS in front of them
ArchiveBox's README and Compose file flag these before you rely on it:
non-negotiableStorage grows fast. The README puts usage between about 1 GB and 50 GB per 1000 snapshots, depending mostly on whether video and audio are saved through yt-dlp, and it advises keeping index.sqlite3 on a local drive. Our snippet keeps everything in the archivebox_data Docker volume, so size the disk before importing a large bookmark export and back that volume up.
non-negotiableArchived JavaScript is untrusted content. The default SERVER_SECURITY_MODE=auto uses isolated subdomains with full replay only on *.localhost, and on an ordinary hostname or IP it serves one domain with replay that has JavaScript turned off. Full replay needs wildcard DNS and TLS and is the safe-subdomains-fullreplay mode.
non-negotiableBy default every archived URL is also submitted to archive.org (ARCHIVEDOTORG_ENABLED=True in the upstream Compose file). Set it to False for a local-only archive, and keep PUBLIC_ADD_VIEW at False unless you want anonymous visitors to queue URLs on your server.
non-negotiableArchiving pages that need a login means giving ArchiveBox a Chrome profile or cookies, and the README points to the wiki's security caveats for that. Pages behind a login can end up in an archive that other users of the instance can read.
non-negotiableSet BASE_URL when you put it behind a domain. Until then the first-run wizard asks for it, and the web and admin interfaces are reached through web. and admin. subdomains on localhost but through one host on a normal IP or domain. Our snippet serves plain HTTP on port 5797, so add TLS before real use.
ArchiveBox used 339 MB of RAM at idle on GCP e2-standard-2 · Ubuntu 26.04 LTS · Docker 29.9.3 (first boot, measured Nov 2026). We list a minimum of 2048 MB.
Is the ArchiveBox install snippet verified?
The ArchiveBox install snippet is marked verified on Ubuntu 26.04 (Nov 2026).