gtsteffaniak/filebrowser

FileBrowser Quantum is a substantial fork of the original FileBrowser project, rebuilt as a self-hostable web file manager with a modern UI, multiple storage sources, SSO logins, and configurable sharing. It is a reasonable candidate if you already self-host services and want a browser-based file layer over directories you control.

The problem

You have files on a home server or VPS, and you want to reach them from a browser rather than setting up SSH or syncing everything to a cloud drive. The usual route is either a heavyweight collaboration suite or a basic directory listing with weak search and no per-user permissions. Quantum changes that step: you point it at one or more local source paths, define users or groups, and get a searchable, thumbnail-rich browser with share links and a text editor. The result is a private file portal you administer through a single config.yaml.

What's different

The main differentiator is source configuration. Instead of one root directory, you can register multiple sources with include/exclude rules, then scope access control to a user or group and a specific source path — useful when different people should see different trees on the same host. Search runs against an index and updates live, with filters for file and folder sizes; thumbnails cover office documents, video, album artwork and 3D models, and folder sizes are shown in listings.

Authentication is broader than the original: OIDC, LDAP, JWT, password with 2FA, and proxy headers are listed as supported login methods. Shares can carry expiration times, anonymous or named access, styling, and separate view/edit/upload permissions. There is also a developer API with long-lived tokens and a Swagger page at /swagger.

The trade-offs are visible in the project's own comparison. The Docker image is listed at 180 MB with ffmpeg and minimum memory at 512 MB, notably heavier than the original's 31 MB image and 128 MB minimum — thumbnail generation and indexing cost resources. Shell commands were deliberately removed from this fork and the author says they will not return, so do not adopt it expecting the original's command execution. S3 and FTP are not supported. Several listed features — custom jobs, tags, trash, metrics, event notifications, archive browsing, document/video/photo conversion — are marked as under construction, not available.

Who should try it

Good fit: someone running a home lab or small team server who wants a browser UI over local directories, with separate accounts, group-scoped visibility, share links, and login through an existing identity provider. The multi-source and per-path access model matters most when several people share one machine but should not see each other's files.

Poor fit: anyone who needs S3 or FTP backends, a mobile app (listed as absent), file comments, or the original's shell command feature. If you mainly need object storage browsing, this project's filesystem-only model rules it out. Those with less than roughly 512 MB to spare, or who want minimal image footprint, should weigh the original FileBrowser instead. If you are migrating from v1, the project points to a migration discussion for v2.0.0, so plan that transition rather than assuming a drop-in upgrade.

Requirements and limitations

It runs as a standalone binary or Docker image on Linux, Windows and macOS. You supply filesystem paths and write a config.yaml; there is getting-started documentation and a separate sources configuration guide. Setup is modest, but ongoing upkeep includes index and thumbnail processing, which is why the memory floor is higher than the original.

The author's comparison chart is the author's own, so treat feature ticks as claims rather than independent verification. Content-aware search is explicitly absent, and S3/FTP are missing. Features marked under construction should not be planned around. The documentation site is English-only, and the project's own comparison lists no mobile app.

HotRepo's take

HotRepo recorded this repository on the monthly Go board at #18 with +535 stars on 2026-10-28; there is no comparable daily-board history, so persistence of interest cannot be judged from the available evidence. Before adopting, verify the v2 migration path and confirm whether the specific authentication method you need is documented as supported.