VectifyAI / OpenKB

项目摘要

OpenKB is an open-source system that compiles raw documents into a structured, interlinked wiki-style knowledge base using LLMs, powered by PageIndex's vectorless, reasoning-based retrieval. It creates a persistent knowledge foundation where information compounds over time rather than being re-derived on every query, with native multi-modal support and a Skill Factory for distilling agent skills.

为什么值得关注

OpenKB addresses a fundamental limitation of traditional RAG—knowledge rediscovery on every query—by compiling documents into a persistent, cross-referenced wiki that accumulates and synthesizes information. Its vectorless retrieval approach eliminates the need for a separate vector database while handling long documents and multi-modal content natively.

优势

• Vectorless reasoning-based retrieval eliminates dependency on external vector databases
• Persistent wiki compilation enables knowledge compounding and cross-referencing over time
• Native multi-modal support for figures, tables, and images alongside text
• Broad document format support including PDF, Word, PowerPoint, Excel, HTML, and URLs
• Skill Factory enables distillation of redistributable agent skills from compiled knowledge

局限性

• No Docker support for containerized deployment
• Relatively new project with limited production track record
• Requires LLM API access for document compilation and maintenance

使用场景

• Enterprise knowledge management and document intelligence
• Building domain-specific AI assistants with persistent memory
• Research literature analysis and synthesis across large document collections
• Creating structured knowledge bases for multi-agent systems