📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG

MCP server for PageIndex. PageIndex is a vectorless reasoning-based RAG system which uses multi-step reasoning and tree search to retrieve information like a human expert would.

Vectorless, Reasoning-Based Retrieval-Augmented Generation (RAG)

📈 FinanceBench evaluation of Mafin 2.5 (Powered by PageIndex)

PageIndex Web是基于 PageIndex 开源项目构建的 Web 可视化RAG系统，提供了完整的文档处理、树结构可视化和智能问答功能。

No description available

No description available

API DEEPSEEKFIED FROM https://github.com/VectifyAI/PageIndex

官方PageIndex的魔改，支持直接定位问题答案所在文档的某个段落

HelloData分析： 1、HelloData.FrameWork:为数据库底层框架，支持多种数据库操作，加入了BaseEntity与BaseLogic，BaseManager两个有关业务逻辑的继承方式。在数据库生成model的时候使用T4生成， 并且生成的数据库表对应的对象类为部分类（partial），如果需要扩展加入当前对象的多个部分类即可。这样做的好处是将数据库生成 的类与业务间的操作分离。BaseLogic的继承BaseLogic<T>，T为操作数据库表对象的泛型，里面包含的常用的新增，删除，修改，获取一个实体，获取实体list ,BaseManager<T, TU>,T为操作逻辑类，TU为操作逻辑对象类。继承后当前操作逻辑类为全局唯一实例，使用了单一模式，操作方法也是包含了那些常用的逻辑操作。 书写Demo: using (DeleteAction delete = new DeleteAction(Entity)) { delete.SqlWhere(cms_user.Columns.id, "1,2,3,4,5", RelationEnum.In); delete.Excute(); return delete.ReturnCode; }； using (UpdateAction update = new UpdateAction(Entity)) { update.SqlKeyValue(cms_user.Columns.createtime, null); update.SqlKeyValue(cms_user.Columns.password, "123456123"); update.Excute(); return update.ReturnCode; }； using (SelectAction select = new SelectAction(Entity)) { if (!string.IsNullOrEmpty(username)) select.SqlWhere(cms_user.Columns.username, username, RelationEnum.Like, ConditionEnum.Or); select.SqlPageParms(pageSize); return select.QueryPage<cms_user>(pageIndex); }； using (SelectAction action = new SelectAction("")) { action.SqlWhere(cms_user.Columns.username, "admin"); action.SqlWhere(cms_user.Columns.password, "123456"); PageList<cms_user> lists= action.QueryPage<cms_user>(1); return null; } 操作数据库的时候可以加入缓存，缓存现支持webcache,Redis，MemberCache ，后两种可以支持分布式部署操作； 2、HelloData.FWCommon：包含加密解密；导出操作：txt,csv，excel；序列化与反序列化：二进制，json，soap,xml； 其他的常用操作，例如：html操作，socket网络爬虫等。 3、HelloData.FWExtend：这个为开发人员项目操作的，基于HelloData.FrameWork的扩展； 4、HelloData.Web：里面加入url重写，多语言模块，ajax请求类似mvc的操作。 操作demo: function ajaxdemo() { $.ajax({ type: 'POST', url: "ajax/demo/do", data: "{ 'result':{ 'Result':-1,'Message':'不支持GET请求','PostTime':'2012-2-2'},'ido':233}", contentType: "application/json", dataType: "json" }); }

No description available

Benchmark 7 retrieval strategies on your own docs — naive vector, contextual, QnA pairs, knowledge graph, RAPTOR, PageIndex, and hybrid. Find which KB architecture fits your data.

A Simple PageIndex Implementation and Agent UI Design

No description available

Hierarchical document indexing for RAG pipelines, written in Rust with Python bindings

A production-ready, Neo4j Graph Database implementation of the PageIndex Vectorless RAG architecture.

Ziglang eXtensiable Builder for SQL or JSON, zig version, sql or json query builder, extensible custom for any database, for any orm framework

No description available

Local Hindsight + PageIndex runtime scaffold (no hosted vector DB)

Independent benchmark of PageIndex tree-based RAG vs Google Gemini, CustomGPT, and OpenAI RAG on SimpleQA-Verified (100 questions, 1000 documents)

Vector RAG vs PageIndex: Side-by-side comparison on Pharmacovigilance documents

TypeScript reimplementation of PageIndex - vectorless, reasoning-based document search using LLM traversal. MIT-licensed alternative to Python PageIndex for Node.js applications.

Block Storage based on Hadoop Distributed File System (HDFS)

Atlas - Enterprise document indexing plugin for OpenClaw. Vectorless RAG using PageIndex with async indexing, incremental updates, and smart caching. Scales from 10 to 5000+ documents. Perfect for financial reports, legal docs, technical manuals, and research papers.

Vectorless, reasoning-based RAG using hierarchical document indexing with Vertex AI. No embeddings, no vector DB - just LLM reasoning.

📄 页索引：基于推理的 RAG 文档索引系统

Go rewrite of PageIndex for generating a hierarchical, reasoning-friendly tree index from PDF documents

PageIndex Implementation : Building advanced vectorless, reasoning-based RAG for long document analysis. No chunking, no vectors—pure LLM tree navigation

A microservice RAG system using PageIndex's reasoning-based approach to achieve high retrieva accuracy without vector embeddings. Built with 9 FastAPI services and Next.js, eliminating theneed for vector databases.

OpenNoteVision LM solves Opennotebook's scanned PDF limitation. Combines Deepseek-OCR-2 visual understading with PageIndex reasoning RAG. Open-source,local deployement but likes NoteBookLM ! Turns your scanned docs into conversational knowlege !