WeKnora: Tencent's Open-Source Enterprise LLM Knowledge Platform

Tencent's WeChat team open-sourced WeKnora, an enterprise LLM knowledge platform that converts scattered documents into three knowledge assets: RAG, ReAct Agent, and Wiki. The platform emphasizes private deployment and full data control, with enterprise features like multi-tenancy and RBAC.
Key insights
Three Core Knowledge Assets
WeKnora transforms raw documents into three types of knowledge assets: a queryable RAG knowledge base, an autonomous reasoning ReAct Agent, and an evolving automatic Wiki. The Wiki mode is highlighted as the most differentiating capability, enabling structured knowledge reproduction beyond simple Q&A.
Enterprise-Grade Features and Security
The platform includes multi-tenancy with four-level RBAC (Owner, Admin, Contributor, Viewer), AES-256-GCM encryption, TLS for gRPC/Redis, SSRF-hardened HTTP client, OIDC JWKS verification, and sandbox isolation. Audit logs are tenant-level. The knowledge base is managed like a code repository with versioning, diff, rollback, and automatic re-indexing.
RAG and Agent Capabilities
WeKnora supports multiple retrieval strategies including BM25 sparse, dense vectors, GraphRAG, and parent-child chunking with pgvector as default vector store. The ReAct Agent can orchestrate knwoledge retrieval, MCP tools, and web search from 11 sources, with a persistent sandbox for safety.