FruxonDocs
ApiKnowledge bases

Updates a knowledge base's settings, including how its articles are embedded for search.

Name, description, locale, automatic publishing and consolidation settings are authoring metadata: changing them does not republish anything. `embedding` (provider, model, chunk size, overlap) and `embeddingLLMConfigId` are different: a new model or chunking cannot reuse a stored vector, so the backing index is cleared and every published article is re-embedded and re-published in this same request, and the response reports the base as publishing. Search returns progressively more of the corpus until that finishes. Only members that differ from the live settings count as a change; a new credential alone re-embeds nothing. The change is refused with 409 while a publish is still running, and it clears the consolidation calibration, whose threshold was measured under the old model.

PATCH
/v1/tenants/{tenant}/knowledgeBases/{knowledgeBase}

Authorization

Bearer
AuthorizationBearer <token>

JWT Authorization header using the Bearer scheme. Enter 'Bearer' [space] and then your token.

In: header

Path Parameters

knowledgeBase*string

The knowledge base id.

Formatuuid
tenant*string

The tenant identifier

Request Body

New metadata values. Null properties are left unchanged.

displayName?string|null
description?string|null
locale?string|null
autoPublish?boolean|null

Set to switch automatic publishing on or off; omit to leave it as it is.

consolidatorAgentId?string|null

Set to change which agent a consolidation pass runs; an empty string clears it, so passes go back to report-only. Omit to leave it as it is.

consolidationPolicy?

Patch the per-base consolidation defaults. Omit the object or any member to leave it unchanged.

embedding?

Changes how the base's articles are embedded for search — provider, model, chunk size or overlap — and re-publishes every document under the new settings in the same request. Members left null keep their current value; a value equal to the current one is not a change.

A new model or chunking cannot reuse a single stored vector, so the index is cleared and refilled: search returns progressively more of the corpus until the re-publish finishes. It is refused while a publish is already running, and it clears the consolidation calibration, whose similarity threshold was measured under the old model.

embeddingLLMConfigId?string|null

The tenant LLM config whose credential embeds the base's articles — required when the new provider needs a different credential (an OpenAI key cannot embed with Vertex AI). Resolved server-side, so the secret never travels in the request. Spelled as on KnowledgeBases.CreateKnowledgeBaseRequest.

Formatuuid
[key: string]?never

Response Body

curl -X PATCH "https://api.fruxon.com/v1/tenants/string/knowledgeBases/497f6eca-6276-4993-bfeb-53cbbbba6f08" \  -H "Content-Type: application/json" \  -d '{}'
{
  "id": "00000000-0000-0000-0000-000000000000",
  "displayName": "string",
  "description": "string",
  "locale": "string",
  "suggestedLocale": {
    "locale": "string",
    "matching": 0,
    "considered": 0,
    "detected": false
  },
  "networkId": "00000000-0000-0000-0000-000000000000",
  "autoPublish": false,
  "automaticPublishWithheld": false,
  "consolidatorAgentId": "string",
  "correctionsAvailable": false,
  "consolidationPolicy": {
    "version": 0,
    "cadence": "MANUAL",
    "mode": "REPORT_ONLY",
    "similarityThreshold": 0,
    "maxClusterSize": 0,
    "recentWriteWindowMinutes": 0,
    "autoAskDailyLimit": 0,
    "calibration": {
      "runId": "00000000-0000-0000-0000-000000000000",
      "similarityThreshold": 0,
      "embeddingModel": "string",
      "locale": "string",
      "calibratedBy": "string",
      "calibratedAt": 0
    }
  },
  "consolidationRecipe": {
    "domain": "string",
    "locale": "string",
    "categories": [
      "string"
    ],
    "provider": {
      "providerId": "string",
      "model": "string",
      "parameters": {},
      "fallbacks": [],
      "builtInTools": [],
      "tenantConfigId": "00000000-0000-0000-0000-000000000000"
    },
    "recipeVersion": 0,
    "upgradeAvailable": {
      "from": 0,
      "to": 0,
      "kind": "AGENTS_ONLY"
    }
  },
  "lastConsolidatedAt": 0,
  "backingAssetId": "string",
  "embedding": {
    "provider": "string",
    "model": "string",
    "chunkSize": 0,
    "chunkOverlap": 0
  },
  "status": "ACTIVE",
  "lastPublishedAt": 0,
  "contentRevision": 0,
  "publishedContentRevision": 0,
  "changesPending": false,
  "publishPendingSince": 0,
  "publishState": "NONE",
  "lastPublishError": "string",
  "lastPublishErrorCategory": "CREDENTIALS",
  "createdAt": 0,
  "modifiedAt": 0
}
{
  "type": "string",
  "title": "string",
  "status": 0,
  "detail": "string",
  "instance": "string",
  "property1": null,
  "property2": null
}
Empty
Empty
{
  "type": "string",
  "title": "string",
  "status": 0,
  "detail": "string",
  "instance": "string",
  "property1": null,
  "property2": null
}
{
  "type": "string",
  "title": "string",
  "status": 0,
  "detail": "string",
  "instance": "string",
  "property1": null,
  "property2": null
}