Reference

Every reference page, grouped by what you are trying to do. If you are new here, start with Lists — a list holds your records, and everything else in Search Stack is built on one. Prefer to search? The box in the sidebar covers every page below, section by section.

Start here

• Lists — what a list is, and the path from creating one to searching it.
• Fields — searchable, facet, coordinates and resource fields, and what each lets search do.
• Searching — matching on words and on meaning, scores, and how results are ordered.
• Groups — search several lists as one, with merged results.

Keeping data in sync

• Feeds — what a feed is: a source we check on a schedule, save as a file, and import into a list.
• Website Feed — point it at a section of your site and every page under it becomes searchable.
• RSS Feed — follow an RSS feed and make each new item searchable as it is published.
• API Feed — pull records from your own API on a schedule.
• Email Feed — forward mail to an address we give you and every message becomes searchable.
• Media Store Feed — drop files in a store and have them imported and re-imported automatically.
• Media Stores — where your files live, the file editor, and sync badges.
• Cleaning — tidy incoming records before they are made searchable.
• Gates — rules that check, rewrite, hold back or refuse a record on the way in.
• Retention — when records expire, and how long the expired ones are kept before they are deleted for good.

Search features

• Query Optimisation — treat a postcode, barcode or order number in a search as a filter rather than as words.
• Synonyms — teach a list that your customers’ words and your catalogue’s words mean the same thing.
• Suggestions — suggestions as you type, drawn from your own data.
• User Generated Suggestions — suggestions that improve as people search.
• Image Search — search by a picture instead of words.
• Playground — try queries against a list and see what ranking does.
• Components — the ready-made search box for your website: install, options, events, styling.

Measuring quality

• Evaluations — turn “is my search any good?” into a number you can watch, so you notice when it gets worse.
• Judges — have a model score results against your own instructions.
• Rerankers — bring your own reranker (a model that reorders results) and attach it to a group.
• Embedding Models — bring your own embedding model (the model that lets search match on meaning); changes apply in the background.

Running it in production

• Search Servers — the engine behind your lists, shared versus per-account indexes, regions, and moving between them.
• Caches — recent answers reused, on the built-in cache every list gets.
• API Versioning — what moves a list to a new version, pinning to one, moving to a newer one, and restoring an older one.
• Telemetry Destinations — send traces, metrics and logs to your own monitoring tools.
• Webhooks — get a message at your URL when something happens, signed so you can trust it and retried if it fails.
• Account Health — is anything wrong right now: a feed that stopped, an import that failed, a service of yours that is not answering.

Accounts and access

• API Keys — create a key, choose its permissions, replace it with a new one.
• Credentials — store your own provider keys (your OpenAI key, say) once and reuse them, encrypted when stored.
• Contributors — give teammates access to a list, and what they can and cannot do.