knowlp-rag

Managing Markdown notes often suffers from inefficient retrieval, as traditional tools like grep or naive vector stores lack contextual understanding, yielding redundant or irrelevant results. The KnowLP-RAG plugin addresses this by transforming notes into a dynamic knowledge graph through dual-graph retrieval, delivering precise reading paths and what you should read next.

Its core capabilities feature a dual-graph retrieval system: the P-Agent handles dependency chains to reveal logical connections between notes; the S-Agent offers similar substitutes, broadening search dimensions. A decay-based forgetting mechanism adjusts note weights based on usage—frequently accessed notes strengthen while unused ones decay, mimicking the "use it or lose it" learning principle. It supports Chinese notes out-of-the-box, requires no GPU, and improves via user feedback loops.

This plugin is ideal for knowledge workers who regularly use Markdown for documentation, such as developers, researchers, or content creators. Whether managing project docs, academic notes, or personal knowledge bases, it provides intelligent retrieval and substitution suggestions, minimizing search time and enhancing knowledge coherence. It's a lightweight yet powerful solution for teams or individuals seeking efficient knowledge management tools.