Recherche Documentation
Hybrid search in the knowledge library combining vector, lexical, and graph search. More comprehensive than pure semantic search.
Parameters
| Parameter | Type | Required | Variable | Description |
|---|---|---|---|---|
query | dynamic value | Yes | Yes | Search text for the knowledge base. |
limit | number | No | No | Maximum number of documents to return (1 to 50). (Default: 10, min 1, max 50) |
tags | array | No | No | Filter documents by tags (e.g. finance, legal). |
fileTypes | array | No | No | Filter by file types (pdf, docx, md, txt, html). |
category | text | No | No | Filter documents by category. |
outputVariable | text | No | No | Output variable name containing the found documents. |
Parameters marked Variable = Yes accept the
{{blockName.field}}syntax.
Output
Output variable : docResults
{
"results": [],
"count": 0
}
Example
Search internal documentation.
Input :
{"query": "procedure de validation fournisseur"}
Output :
{"results": [{"id": "doc-42", "title": "Validation fournisseur", "content": "...", "score": 0.88}], "count": 1}
Tip
Prefer this block over semantic_query for complex searches. RRF fusion combines 4 sources for better recall.