Skip to main content
Use search when you already have indexed content and want immediate ranked retrieval.

​
Request

curl "$BULKGRID_BASE_URL/api/v1/search" \
  -H 'Content-Type: application/json' \
  -H "x-api-key: $BULKGRID_API_KEY" \
  -d '{
    "query": "pricing page",
    "limit": 3
  }'

​
Important request fields

  • query: the search text
  • limit: number of results to return, up to 50
  • collectionId: optional collection scope
  • fullTextWeight, semanticWeight, rrfK: ranking controls

​
Example response

{
  "query": "pricing page",
  "scoring": {
    "fullTextWeight": 0.3,
    "semanticWeight": 0.7,
    "rrfK": 60
  },
  "results": [
    {
      "chunk_id": "3f7a3578-7cec-4722-bab4-b37ba0b9c2ab",
      "source_document_id": "0866c8e8-a384-4f52-b83f-327a1afecf1d",
      "source_id": "a3c05cb9-5f19-4106-b8b6-b255c36c0f6d",
      "url": "https://docs.example.com/pricing",
      "title": "Pricing",
      "chunk_index": 0,
      "text_content": "Choose the plan that matches your usage...",
      "metadata": {},
      "fts_rank": 0.81,
      "semantic_rank": 0.94,
      "score": 0.9
    }
  ],
  "count": 1
}

​
What this tells you

  • search is synchronous
  • each result points to a source URL and source document identifiers
  • ranking combines lexical and semantic signals
  • you can use the response directly in retrieval and answer-generation workflows

​
Next step

Read the fuller Search.