Search API

Query across all platform entities with full-text, keyword, or semantic search. Search covers catalog records, operational entities, comments, document pages, and Data Question threads when those features are enabled and indexed.

Global Search

GET /search

Query Parameters

Parameter	Type	Required	Description
q	string	Yes	Search query
space_id	uuid	Yes	Scope to specific space
mode	string	No	Search strategy: auto (default), keyword, or semantic
entity_types	string	No	Comma-separated entity types such as products,issues,meetings,users,documents,data_questions
limit	integer	No	Max results per type (default: 10)

Example Request

curl -X GET "https://api.qarion.com/search?q=customer&space_id=SPACE_UUID&mode=auto" \
  -H "Authorization: Bearer YOUR_API_KEY"

Response

Results are grouped by entity type:

{
  "products": [
    {
      "id": "...",
      "name": "Customer Events",
      "description": "Raw event stream...",
      "product_type": "table",
      "score": 0.95
    }
  ],
  "issues": [
    {
      "id": "...",
      "title": "Customer ID nulls",
      "status": "open",
      "score": 0.82
    }
  ],
  "meetings": [],
  "users": [
    {
      "id": "...",
      "name": "Customer Data Team",
      "email": "customer-team@example.com",
      "score": 0.75
    }
  ]
}

Search Modes

The mode parameter controls the search strategy used:

Mode	Description
auto	(default) Uses keyword search and automatically adds semantic results when AI embeddings are configured
keyword	Full-text keyword search only (Elasticsearch with PostgreSQL fallback)
semantic	Vector-similarity search using AI embeddings — requires an embedding backend to be configured

Keyword Search

curl -X GET "https://api.qarion.com/search?q=revenue+metrics&space_id=...&mode=keyword" \
  -H "Authorization: Bearer YOUR_API_KEY"

Keyword search uses Elasticsearch fuzzy matching with field-weighted scoring:

Field	Weight	Description
Name/Title	High	Primary identifier
Description	Medium	Summary text
Tags	Medium	Category labels
Documentation	Low	Long-form content

Published document pages are indexed after publish or tree metadata updates and removed after deletion. Data Question threads are reindexed after thread or message changes so search results stay aligned with the latest discussion.

Fuzzy matching handles typos: custmer → customer, anlytics → analytics.

Semantic Search

curl -X GET "https://api.qarion.com/search?q=tables+about+customer+spending&space_id=...&mode=semantic" \
  -H "Authorization: Bearer YOUR_API_KEY"

Semantic search converts the query into a vector embedding and finds products whose metadata vectors are closest in meaning. This is useful for intent-based queries where exact keywords may not appear in the data.

Semantic search requires an embedding backend to be configured. Qarion defaults to the local backend (sentence-transformers) which works out of the box.

Highlighting

Keyword search results include highlighted matches (when available):

{
  "name": "<em>Customer</em> Events",
  "description": "Raw event stream for <em>customer</em> tracking"
}

Entity-Specific Search

Products Only

GET /search?q=metrics&space_id=...&entity_types=products

Multiple Types

GET /search?q=order&space_id=...&entity_types=products,issues

Search Performance

• Results are cached for 60 seconds
• Maximum 50 results per entity type
• For large result sets, use pagination via the catalog API

Related Endpoints

• Catalog API — For filtered product listing with pagination
• Issues API — For filtered issue queries
• Documents API — For document-page lifecycle and indexing behavior
• Data Questions API — For searchable Q&A threads