Platform
Knowledge Collections
Organize documents into collections, scope retrieval per collection, and attach them to one or many domains.
Knowledge Collections group the documents that back retrieval — each collection has its own retrieval config (embedding model, chunking strategy) and can be attached to one or many domains, so you don't have to re-upload the same source library for every domain that needs it.
API reference
GET /dashboard/collections # list collections
POST /dashboard/collections # create a collection
PATCH /dashboard/collections/:id # update config
DELETE /dashboard/collections/:id # delete
POST /dashboard/collections/:id/domains/:domainKey # attach to a domain
DELETE /dashboard/collections/:id/domains/:domainKey # detach from a domain
GET /dashboard/collections/:id/documents # list documents in the collection
POST /dashboard/collections/:id/documents/:documentId # attach a document
DELETE /dashboard/collections/:id/documents/:documentId # detach a document
GET /dashboard/collections/:id/analytics # retrieval hit-rate, top queries
GET /dashboard/collections/:id/connections # which domains use this collectionCreate a collection:
POST https://api.liyaengine.ai/dashboard/collections
Authorization: Bearer <jwt>
Content-Type: application/json
{
"name": "Support Knowledge Base",
"tags": ["support", "policies"],
"visibility": "workspace",
"defaultEmbeddingModel": "text-embedding-3-large",
"defaultChunkingStrategy": "semantic"
}Uploading documents
Documents are uploaded independently, then attached to a collection:
POST /dashboard/knowledge/documents # upload a document
GET /dashboard/knowledge/documents # list documents
GET /dashboard/knowledge/documents/:id # get one
DELETE /dashboard/knowledge/documents/:id # deleteOnce attached to a collection, a document is retrievable by every domain the collection is connected to — update the source once, every domain sees the change.