Coming soon

The Lookeen RAG API and MCP integration are not yet available. The information below describes planned functionality.

Lookeen RAG API

The retrieval API returns ranked Lookeen passages for grounding an agent or language model. It does not generate an answer or invoke an LLM. This page renders the GET /v1/rag contract as HTML.

Endpoint

URI	/v1/rag
Method	POST
Content type	application/json
Retrieval	lexical
Authentication	Authorization: Bearer <Lookeen API key>

Request body

Field	Type	Description
query	String	Lookeen query to execute. Required and non-blank.
topK	Integer	Maximum passages to return. Default: 10, minimum: 1, maximum: 100.
userId	Integer	Optional user ID whose document permissions are applied.
groupIds	Array<Integer>	Optional group IDs whose document permissions are applied.
filters	Array	Optional structured document filters containing column, op and value.

Filters

Field	Description
column	Searchable Lookeen column name; required.
op	Filter operator; required.
value	Filter value; optional only for operators such as EMPTY and NOT_EMPTY.

Supported operators: IN, NOT_IN, EQUALS, LARGER, NOT_EQUALS, LIKE, NOT_LIKE, SMALLER, STARTS_WITH, DOES_NOT_START_WITH, DOES_NOT_END_WITH, ENDS_WITH, NOT_EMPTY, EMPTY, RANGE, NOOP, CONTAINS and COMPLEX.

Example request

curl -X POST https://lookeen.example/v1/rag \
  -H 'Authorization: Bearer <api-key>' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "Project Phoenix risks",
    "topK": 5,
    "userId": 42,
    "groupIds": [3, 8],
    "filters": [
      {"column": "extension", "op": "EQUALS", "value": "pdf"}
    ]
  }'

Response

Field	Description
query	Normalized submitted query.
retrieval	Retrieval strategy used; currently lexical.
totalMatches	Total matching documents before topK is applied.
results	Results ordered by decreasing relevance. One passage is returned per matching document.
elapsedMs	Server-side retrieval time in milliseconds.

One results item

Field	Description
rank	One-based result rank.
score	Relevance relative to the first result, from 0 to 1; omitted when unavailable.
documentId	Stable Lookeen document identifier for citation and later lookup.
title	Document title.
path	Document path, when available.
url	Document URL, when available.
mimeType	Document MIME type, when available.
itemType	Lower-case Lookeen item type, when available.
modified	ISO-8601 modification time, when available.
passage	Plain-text passage selected for this query and suitable for grounding.

Example response

{
  "query": "Project Phoenix risks",
  "retrieval": "lexical",
  "totalMatches": 18,
  "results": [
    {
      "rank": 1,
      "score": 1.0,
      "documentId": "project-status~12~42",
      "title": "Project status Q3",
      "path": "S:\\Projects\\Phoenix\\Project-status_Q3.pdf",
      "mimeType": "application/pdf",
      "itemType": "file",
      "modified": "2026-10-14T08:30:00Z",
      "passage": "The API release is delayed. Two data protection reviews remain open."
    }
  ],
  "elapsedMs": 34
}

← Back to Enterprise Search for AI (RAG)