WeKnora
Most teams do not have a knowledge problem. They have a scattered documents problem, and every answer starts with another search. WeKnora, from Tencent, turns raw documents into one system you can query three ways: fast RAG answers with citations, a multi-step agent that can use tools and skills, and a wiki that organises the same knowledge into linked pages. A new upload merges into existing pages, and deleting a document rewrites or removes the pages that cited it. Version 0.9.0 arrived on 29 October 2026 with an in-app assistant, multi-agent collaboration, and automations. It now has 36,779 stars on GitHub and runs self-hosted with Docker, or as a single Lite binary.
What you get
- Three ways to query the same knowledge: cited RAG answers, a tool-using agent, and a linked wiki
- Knowledge that stays consistent: uploads merge into existing pages, deletions rewrite or remove the pages that cited them
- Self-hosted with Docker, or as a single Lite binary
- Version 0.9.0 adds an in-app assistant, multi-agent collaboration, and automations
Start with the documents your team already searches for, and let the wiki grow from what people actually ask.
Repo: https://github.com/Tencent/WeKnora
Quick poll
What would you use WeKnora for first?
- Fast RAG answers with citations
- The multi-step agent
- The auto-built wiki
- Self-hosting it with Docker
Vote in the comments.
If you found WeKnora interesting, I write about tools like this every week in The AI-Native Engineering Leader, what actually works for engineering leaders building with AI agents, and what to skip.
Top comments (0)