plugin-knowledge-base
v1.1.2
Published
Provides Knowledge Base management, Vector Store, Vector Database (PGVector), and RAG retrieval capabilities for AI Employees.
Downloads
88
Maintainers
Readme
plugin-knowledge-base
AI Knowledge Base plugin for NocoBase. Provides RAG (Retrieval Augmented Generation) capabilities for AI Employees.
Features
- Knowledge Base Management — Create and manage knowledge bases with document upload
- Access Control (ACL) — 4 access levels: Private, Basic, Shared (role-based), Public
- Shared mode:
allowedRoles(read/search) anduploadRoles(upload documents) per KB - Permission enforcement on all operations: list, create, destroy, reprocess, RAG search
- Shared mode:
- Vector Store — Bind Embedding models with Vector Databases
- Vector Database — PGVector connection management with connection testing
- Document Processing — Automatic document parsing, chunking, and vectorization
- File upload (.pdf, .txt, .md, .doc, .docx, .ppt, .csv, .json)
- Paste text documents directly
- Auto-delete source file option after successful embedding
- RAG Retrieval — Semantic search for AI Employee conversations via Work Context
- KB Management Popup in Chat — Users can browse, upload, and select KBs directly from AI chat
- Left sidebar with KB list and access level badges
- Document table with status, chunk count, and actions
- File upload and drag-and-drop support
- "Select for Chat" toggle for RAG context
Requirements
- NocoBase 2.x
- Plugin AI (
@nocobase/plugin-ai) enabled - PostgreSQL with
pgvectorextension (for PGVector provider)
Installation
yarn pm add plugin-knowledge-base
yarn pm enable plugin-knowledge-base