Search

Vector, keyword, and hybrid search, built on your content models with no separate index to maintain.

Documents embedded automatically as you write

Search modes — vector, keyword, hybrid

Typical reindex time for 1,000 documents

Query shape across all three modes

Search that understands meaning

Vector search on every model, out of the box

Every document is embedded automatically as you write it, so semantic search works from day one — no separate indexing pipeline, and no embeddings to manage by hand.

Embeddings generated and kept in sync automatically
Query by meaning, not just exact keywords
Scoped to a dataset, a model, or your whole project

const results = await cairn.search({
  query: 'how to branch content safely',
  mode: 'hybrid',
  limit: 5,
})

Tune relevance your way

Blend keyword precision with semantic recall

Hybrid search combines exact keyword matching with vector similarity, so acronyms, product names and typos are handled as well as conceptual queries — with weighting you can adjust per use case.

Switch between keyword, vector, and hybrid per query
Boost fields like title or tags in relevance scoring
Filter results by any field on the model

$ cairn search reindex --model article
✓ Re-embedded 1,842 documents in 6.2s

One search API

Three modes, the same query shape

Vector search

Semantic search over embeddings generated automatically for every document.

Keyword search

Fast, precise full-text search for exact terms, names and phrases.

Hybrid ranking

Blend both modes with adjustable weighting to tune relevance per use case.

Add search to your content

Search is on by default — no index to configure before your first query.

npm create cairn View docs