Tencent/WeKnora

What it solves

WeKnora is an enterprise-grade knowledge framework designed to transform scattered documents into a queryable, reasoning-capable knowledge asset. It solves the problem of fragmented information by providing a centralized system for document understanding, semantic retrieval, and autonomous reasoning across various data sources.

How it works

The system uses a modular pipeline that integrates document parsing, vectorization, and LLM inference. It employs three primary modes of operation:

• RAG-based Quick Q&A: Provides fast, accurate answers based on knowledge base lookups.
• ReAct Agent: An autonomous agent that orchestrates retrieval, web search, and MCP (Model Context Protocol) tools to solve complex multi-step tasks.
• Wiki Mode: Agents automatically distill raw documents into a structured, interlinked Markdown knowledge base accompanied by an interactive knowledge graph.

It supports multi-source ingestion (Feishu, Notion, Yuque, RSS) and integrates with a wide array of LLM providers, vector databases, and object storage backends.

Who it’s for

WeKnora is built for enterprises and developers who need a self-hostable, secure knowledge management system with robust access control (RBAC), observability (via Langfuse), and the ability to integrate knowledge agents into existing IM channels (Slack, Telegram, WeCom, etc.) or external websites.

Highlights

• Multi-Source Sync: Auto-syncs knowledge from Notion, Feishu, and Yuque.
• Autonomous Wiki Generation: Automatically creates structured wiki pages and knowledge graphs from raw documents.
• Extensive Integrations: Supports 20+ LLM providers, multiple vector DBs (pgvector, Milvus, Qdrant), and various object storage options.
• Enterprise Security: Features 4-tier RBAC, AES-256-GCM encryption, and scoped API keys.
• Omnichannel Deployment: Deployable via Web UI, Chrome Extension, CLI, and various IM platforms.