VectifyAI/pageindex-mcp

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.

What it solves

PageIndex MCP allows LLMs to interact with long documents (like PDFs) without hitting context window limits or relying on traditional vector databases. It replaces vector similarity search with a reasoning-based approach to information retrieval, mimicking how a human would use a book's index to find specific information.

How it works

Instead of chunking text into vectors, PageIndex creates hierarchical tree structures of documents. LLMs use the Model Context Protocol (MCP) to navigate these trees and perform multi-step reasoning and tree searches to retrieve the relevant passages. This process preserves the full document context and structure, avoiding the need for "top-k" retrieval of isolated fragments.

Who it’s for

Developers using MCP-compatible agents or LLM platforms (such as Claude, Cursor, and Vercel AI SDK) who want to integrate high-accuracy, vectorless RAG into their AI agents.

Highlights

  • Vectorless RAG: Eliminates the need for vector databases and infrastructure overhead.
  • Reasoning-Based Retrieval: Uses tree search and reasoning trajectories for higher accuracy and transparency.
  • No Chunking: Maintains the full context and structure of the document.
  • MCP Integration: Seamlessly connects to agent frameworks like Claude, LangChain, and OpenAI Agents SDK.

Related

  • Project
    VectifyAI/PageIndex

    PageIndex is an open‑source RAG system that replaces vector‑store retrieval with a hierarchical tree index built from a document’s layout. An LLM reasons over this tree to find relevant sections, giving explainable, citation‑ready answers without any vector DB or chunking. The SDK works locally (≈ $0.001 / page indexing cost) or via PageIndex Cloud for OCR‑heavy, large‑scale corpora. Benchmarks show 98.7 % accuracy on FinanceBench and up to 16× lower query cost versus feeding whole PDFs to a model.

  • Project
    docling-project/docling-mcp

    An MCP server that provides AI agents with tools for converting PDFs into structured formats and generating new documents using the Docling library.

  • Project
    openags/paper-search-mcp

    An MCP server for searching and downloading academic papers from multiple open-access and public sources, designed for LLM-enhanced research workflows.

  • Project
    qdrant/mcp-server-qdrant

    An official Model Context Protocol (MCP) server that enables LLMs to use Qdrant as a semantic memory layer for storing and retrieving information.

  • Project
    microsoft/playwright-mcp

    An MCP server that enables LLMs to interact with web pages using structured accessibility snapshots via Playwright, bypassing the need for vision models.