Collections from your data sources
Create a Search Collection from one or more data sources (crawled web pages). Indexing runs in the background and the collection becomes searchable when it's ready.
CatalogLatency Search: typically under a second. Indexing: runs in the background after create / update. Async true Rate Limit Per API key Max Input List endpoints accept up to 200 per page
cn8/search
Build a Search Collection from one or more data sources (crawled web pages), then run
Learn more about Search Text NLPsyncInstant previewsearchragweb
Search is built in your dashboard
Search runs on your own indexed content — create a collection, point it at your sources, and indexing happens in the background. Then query it from anywhere with your key.
1
Create a search collection from your sources
2
Let it index and run test queries in the dashboard
3
Query it from your app via core-search
No account yet? Start free — 100 credits, no card.
Overview
Build a Search Collection from one or more data sources (crawled web pages), then run full-text queries against it. Results come back ranked, with matching snippets you can highlight, plus language and date filters and pagination. A collection can be made public for a set of whitelisted domains — so you can embed search on your own site without exposing an API key — or kept private behind your key. Indexing runs in the background; you create or update a collection and it's searchable once it's ready.
Key capabilities
Full-text search with filters
Query by keyword with ranked results, optional snippet highlighting, language and date filters, and pagination.
Public or private access
Whitelist domains to allow public, key-less search from your own site, or keep a collection private behind your API key.
Query history
Review the queries run against a collection, newest first, with date filters.
When to use it
On-site search
Index your website or docs and add a fast, ranked search box — public on your own domains, no key in the browser.
Knowledge search
Search across crawled internal pages or a curated set of sources from your backend.
Research and monitoring
Query a collection with date filters to find when and where topics appear across your sources.
Input & output
input
A collection built from data_source_ids, and a search query (with optional language / date / pagination filters).
JSON body
output
Ranked hits with title, URL, description and relevance score, plus pagination — and optional per-field highlight fragments.
JSON
Guides & tips
How it works
- 1Create a Search Collection from one or more
data_source_ids(your crawled web pages). - 2Indexing runs in the background — poll Get Collection Details until
statusisready. - 3Search the collection with a
query, optionally filtering by language and date and turning onhighlight. - 4Update the data sources (or call Reindex) whenever your content changes.
Public vs private access
- Add
public_domainsto a collection to allow public, key-less search from those domains — ideal for an on-site search box where you don't want an API key in the browser. - Without
public_domains, searching requires a valid API key. Query history is always private to the collection's owner key.
Keeping a collection fresh
- Changing a collection's
data_source_idsre-indexes it automatically. - To rebuild against the current sources without changing them, call Reindex (or update with
force_reindex: true).
Specs
Schema
Request body
queryrequiredstringThe search query.
languagestringRestrict results to a language (ISO 639-1).
date_fromstringOnly include pages from this date onward (ISO date).
date_tostringOnly include pages up to this date (ISO date).
pageintegerPage number.
page_sizeintegerResults per page.
highlightbooleanWhen true, results include highlighted snippets of the matched text.
Pricing
Search Collections are free to create, index, search and manage.
| Service | Unit | Price |
|---|---|---|
| All Search endpoints | item | Free |
Creating, updating, re-indexing, searching and reading history are all free.