Documentation
These pages are for the people who run Digital Collections Explorer: the person at a gallery, library, archive, or museum who sets up a site, and the developer who extends it. The README covers installing it locally in a few steps.
Setting up a collection
Section titled “Setting up a collection”- Indexing large or remote collections: profile a collection first, then index it from a folder, an S3 bucket or a manifest, with Python or Docker.
- Mapping your data: bring a catalog into a collection, from a spreadsheet to a museum’s collection system, and what each part of the site needs.
- Collection format: the folder a site serves and every table in it, and sites that hold the collections of several institutions.
- Kinds of material: images are always shown; 3D models, PDFs and deep zoom are options.
- Describing a collection to visitors: the title, suggested searches, license and links, and the link back to each record.
- Searching by appearance and by catalog: how the two channels are weighed and what each result explains.
- Models: SigLIP, SigLIP 2 or CLIP, and pinning a version.
- Limits that protect a public site: uploads, searches per visitor and searches at once.
Deploying
Section titled “Deploying”- On a Linux server: one command that keeps a site running across reboots.
- On AWS: one command that shows its cost first, then HTTPS, a domain of your own, updates and prepared collections.
- Under a path on an existing website: serve a site at
https://library.example.org/maps/rather than on a domain of its own.
Extending
Section titled “Extending”- Making a frontend: an interface of your own, in its own repository.
- MCP server: let Claude, ChatGPT or an editor’s agent search a site, with the tools it offers and the limits that apply.