Skip to Content
Docs are being rebuilt — start at Introduction → How it works.
API referenceKnowledge bases APIKnowledge bases API

Knowledge bases API

A knowledge base is a searchable collection of content (text, web pages, or files) that agents — or your own code — can retrieve from. The API splits into two pages by access level:

PageAccessWhat it covers
Manage knowledge basesAdmin keyCreate, list, get, update, delete; append content; upload files.
Search & queryMember keySemantic search, RAG question-answering, ingestion status, embeddings.

All endpoints are under https://tmmate.ai/api/v1/knowledge-bases and return the { "data": ... } envelope.

Knowledge base types

A knowledge base has one of three source types, chosen at creation:

  • text — raw text content you supply.
  • website — URLs that are fetched, cleaned, chunked, and embedded.
  • files — an empty shell you then attach files to (PDF, DOCX, DOC, TXT, CSV, Markdown).

Content is chunked and embedded with openai/text-embedding-3-small (1536 dimensions). Embedding is synchronous — content is searchable as soon as the create/append/upload call returns 201.

KnowledgeBaseResponse

The object returned by the management endpoints:

FieldType
idUUID
workspaceIdUUID
titlestring
descriptionstring | null
typestring
createdByUUID | null
createdAt / updatedAtISO 8601 | null
contentSizeobject | null — { bytes, formatted, wordCount, chunks, fileCount, webPageCount }