Curator

Curator is a self-hosted photo gallery CMS written in Go. A local admin UI manages galleries, photos, metadata, access, and publishing; the public site is rendered as static HTML and images, with access-control configuration when needed.

Curator supports hierarchical grid and story galleries, unlisted and protected content, simple photo tags, automatic image resizing, EXIF facets, Atom feeds, embedded themes, and publishing from Lightroom Classic.

Initial setup and Curator-specific settings live in the admin UI. Galleries and photos can then be managed there or published day to day from Lightroom Classic. The admin supports gallery hierarchies, contact-sheet review and ordering, photo metadata overrides, source-aware tag review and correction, block-based story editing, presentation defaults, access users, facets, themes, and publishing.

Photo tags can be entered in Curator or imported from embedded and sidecar XMP, IPTC keywords, and Lightroom Classic. Curator tracks each assignment source so metadata refreshes and Lightroom republishes do not erase tags entered in the admin.

Screenshots

Quick start

Curator requires Go 1.26 or newer.

git clone https://github.com/tkjaer/curator.git
cd curator
go build -o curator .

./curator serve -content ./site

Open http://127.0.0.1:8080/. Curator creates the content root and database on first launch, then the admin UI guides the rest of the setup. The content root contains the SQLite database and original images; it is the source of truth and should be backed up.

site/
├── cms.db
├── originals/
└── output/

Set an admin password before exposing the admin through a reverse proxy:

./curator set-password -content ./site

CLI and deployment

Curator can run on the server, with the web server serving its generated output directory directly. It can also run entirely on a local computer and publish only the generated output to a remote static host with rsync; the remote host does not need Curator or access to the content database and originals.

The CLI is useful for automation and headless deployments. Publishing can also be started from the admin UI. Configure an rsync destination under Settings → Publishing and Publish site will build locally, then deploy the generated output after a successful build. The machine running Curator needs rsync and working SSH access to the destination.

Create a content root explicitly when not starting with serve:

./curator init -content ./site

Import a folder of images as a new gallery:

./curator import -content ./site -title "Summer trip" /path/to/photos

Re-read source EXIF and XMP metadata while preserving manual overrides:

./curator rescan -content ./site

Build the public site into site/output:

./curator build -content ./site

The output directory is disposable and can be served by any static web server. To build and copy it to a remote host with rsync:

./curator publish -content ./site -target user@example.com:/srv/www/photos

Before enabling remote deletion, use Settings → Publishing → Preview remote changes to see itemized uploads, updates, and stale files without changing the destination. The equivalent CLI dry run is:

./curator publish -content ./site -target user@example.com:/srv/www/photos -no-build -dry-run -delete

To generate the Atom feed, configure the public base URL under Settings → Site, then enable Generate /feed.xml under Settings → Publishing.

Use ./curator help for all commands and options.

Camera and lens metadata

See the camera and lens metadata guide for replacing scanner names with the camera that exposed the photo, correcting lens names, tagging manual or adapted lenses, and configuring automatic resolution.

Lightroom Classic

The publish-service plugin is in lightroom/Curator.lrplugin. See the Lightroom setup guide for installation and token setup. Once connected, Lightroom can drive the day-to-day publishing workflow: collection sets become parent galleries, published collections become galleries, and publishing synchronizes photos, ordering, metadata, and removals before publishing the site, including rsync deployment when configured.

Development

See docs/development.md for running tests, starting the admin locally, and previewing the generated site. See docs/architecture.md for the data model, generated site layout, access-control setup, and design decisions.