The WeChat AI team has open-sourced version 0.8.0 of WeKnora (Venus), a knowledge management framework. This framework aims to make knowledge bases executable, rather than merely queryable, thereby addressing the challenge of transforming knowledge into verifiable results when deploying large models. WeKnora features a complete pipeline architecture, including anydoc parsing, Wiki-style automatic structuring of knowledge, and GraphRAG optimization for retrieving related questions. Its core Skill Sandbox Runtime can encapsulate knowledge base content into skills, execute them securely in a sandbox, and incorporate long-term memory functions to achieve a full workflow of parsing, organizing, retrieving, executing, and remembering. The architecture is modular, allowing users to select components as needed, and is licensed under MIT for both individual and enterprise users.
