Vector Search
Direct vector similarity search without AI synthesis.
The vector search endpoint performs raw semantic search against your knowledge base embeddings. Unlike the Agent endpoint, it returns matching chunks directly without AI synthesis.
Search
POST /v1/kb/queryScope: kb:read
{
"query": "revenue growth projections",
"workspace_id": "ws_abc123",
"top_k": 5
}Parameters
| Field | Type | Description |
|---|---|---|
query | string | Required. The search query (will be embedded for similarity search) |
workspace_id | string | Required. Workspace to search |
top_k | number | Number of results to return (default: 5) |
Response
{
"results": [
{
"document_id": "doc_abc",
"filename": "Q4-Report.pdf",
"chunk_index": 12,
"content": "Revenue grew 40% year-over-year...",
"score": 0.92,
"metadata": {
"page": 3
}
}
]
}Legacy Endpoint
POST /queryIdentical to POST /v1/kb/query. Maintained for backward compatibility.