Query your knowledge base programmatically using the hybrid search engine. Free plan: 20 semantic searches/day (falls back to keyword-only after limit). Pro and Team: unlimited.
/api/searchExecutes a cross-cutting search against the user's entire archive. Uses indexed PostgreSQL full-text search, exact partial matching, trigram typo-tolerant fuzzy matching, and semantic similarity for longer natural-language queries. Returns up to 10 of the most relevant results.
| Name | Type | Description |
|---|---|---|
queryrequired | string | The natural language query or keyword phrase to search for. |
curl -X POST http://localhost:3000/api/search \
-H "Authorization: Bearer <token>" \
-H "Content-Type: application/json" \
-d '{"query":"PostgreSQL vector indexing strategies"}'{
"results": [
{
"id": "uuid-123",
"url": "https://pgvector.com",
"title": "Scaling pgvector",
"tldr": "How to optimize HNSW indexes...",
"domain": "pgvector.com",
"tags": ["Database", "Performance"],
"keyInsights": ["Insight 1"],
"contentType": "reference",
"status": "ready",
"collectionId": "collection-uuid",
"searchReason": "Tagged with \"Database\"",
"score": 2.45,
"semantic_score": 0.82,
"keyword_score": 0.81,
"partial_score": 0.0,
"fuzzy_score": 0.0
}
],
"semanticLimitReached": false
}