WeKnora

Overview

WeKnora is an open-source, LLM-powered knowledge framework from Tencent for enterprise document understanding, semantic retrieval and reasoning. It brings a team's documents together so they can be searched, reasoned over and kept up to date. RAG is used to look things up, the agent handles multi-step tasks, and the wiki organizes knowledge, with all three working on the same knowledge bases. Version 0.9.0 adds the WeKnora assistant, multi-agent collaboration, automations and execution permissions.

Key Features

• RAG question answering with hybrid semantic and keyword retrieval, multimodal embedding for images, citations that jump to the cited passage in the original document, and an image gallery filterable by source, caption and OCR text.
• Agent task execution with on-demand attachment reading, multi-step reasoning, skills and session-persistent sandboxes on Docker, E2B or Cube, an interactive terminal and graphical desktop, MCP tools, cross-session long-term memory, and operation of the user's own Chrome or Edge
• Automatic wiki that extracts people, products and concepts from documents into linked pages with source references, a knowledge graph with group focus, page editing and version rollback, plus chunk-level editing with revision history.
• WeKnora assistant for product questions and task help, multi-agent collaboration on independent subtasks, one-off or recurring automations, and execution permissions with read-only, confirm-before-change and automatic-write modes.
• Data sources including Feishu wiki and Feishu Drive, Confluence, GitLab, DingTalk Docs, Tencent IMA, Notion, Yuque, Seafile and RSS; access through WeCom, Feishu, Slack, Telegram and other IM apps, a browser extension, an embed widget, a built-in MCP Server, API and CLI; and a
• Enterprise controls including multi-workspace RBAC, per-workspace audit logs, a runtime task-queue dashboard with worker-pool governance, and Langfuse tracing for agent steps, token usage and pipelines.

Use Cases

• Team knowledge Q&A where answers cite the original passage so readers can verify them.
• Multi-step work such as gathering, organizing and cross-checking material using knowledge bases, skills, sandboxes, MCP tools and the local browser.
• Turning existing documents into a browsable, editable wiki with a knowledge graph, source references and version history.
• Reaching the knowledge base from everyday tools such as IM apps, a browser extension, MCP clients like Cursor and Claude, or the API.

Getting Started

• WeKnora can be used online through the WeChat Dialog Open Platform, deployed from a Tencent Cloud Lighthouse application template, or self-hosted. For Docker Compose, install Docker, Docker Compose and Git, then clone the repository, enter the directory, copy .env.example to

Deployment & Requirements

• Docker Compose is the standard deployment with all features and multiple services. Kubernetes with the Helm chart in the helm directory suits production clusters. The Lite single binary targets local or low-resource use with no external dependencies, using SQLite and an

Before You Adopt

• License: Other. Review its terms before using, modifying, or distributing the project.