# VectifyAI/PageIndex

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

Repository: https://github.com/VectifyAI/PageIndex
Canonical: https://ross.abutalabs.com/products/pageindex
Homepage: https://pageindex.ai
Language: Python
License: MIT
License Family: permissive
Topics: agentic-ai, agents, ai, ai-agents, context-engineering, llm, rag, reasoning, retrieval, retrieval-augmented-generation, vector-database, information-retrieval
Last push: 2026-08-26T19:50:00+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 37
- inputs: {"age_days": 519, "days_push": 7, "days_rel": 8, "gap_med": 6, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 35333, forks 3114 (observed 2026-08-28T04:12:00.798797+00:00)

## What it is
PageIndex is a Python SDK and framework for vectorless, reasoning-based RAG that replaces vector similarity search with a hierarchical tree index over documents, which an LLM navigates agentically to retrieve relevant content. It supports local mode with your own LLM key or a hosted cloud API, plus MCP integration, with no embeddings, chunking, or vector database required.

## Use cases
- build RAG over long complex PDFs without a vector database
- retrieve relevant sections from documents by reasoning instead of embeddings
- index PDF documents into a hierarchical tree structure
- add traceable document retrieval to AI agents via MCP
- chat with and analyze long professional documents
- avoid chunking and embedding pipelines for document Q&A

## When to choose
- your documents are long, structured, and require contextual or multi-step reasoning to retrieve accurately
- you want explainable, traceable retrieval grounded in source sections
- you want to avoid maintaining embeddings, chunking, and vector database infrastructure
- you need MCP or API integration for agentic document retrieval

## When to avoid
- you need fast, cheap retrieval over massive corpora where LLM-per-query reasoning cost is prohibitive
- your documents are short or unstructured with no useful hierarchy
- you need sub-second retrieval latency
- you want a fully offline solution without any LLM API key

## Facets
- artifact type: library
- maturity: active
- function: rag, search-engine, llm-inference, agent-framework, pdf, mcp
- domain: large-language-models, pdf
- platform: python, cli, self-hosted, cloud
- tags: vectorless-rag, tree-index, document-retrieval, reasoning-based-retrieval, no-embeddings, document-analysis, retrieval-augmented-generation, natural-language-processing, ai-agents, search

## Member repositories
- VectifyAI/PageIndex (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:00.798797+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:26:58.039018+00:00, confidence not recorded.
  - readme: https://github.com/VectifyAI/PageIndex (fetched 2026-08-28T04:12:00.798797+00:00, sha 192321c884cc)
  - homepage: https://pageindex.ai (fetched 2026-08-29T07:48:01.604134+00:00, sha b2d108e19a06)
  - site_page: https://pageindex.ai/about (fetched 2026-08-29T07:48:01.607877+00:00, sha 2fee92fee4d0)
  - site_page: https://docs.pageindex.ai (fetched 2026-08-29T07:48:01.609635+00:00, sha 3ef3ed6e22e1)
  - registry_pypi: https://pypi.org/pypi/pageindex/json (fetched 2026-08-29T07:48:01.611521+00:00, sha 8bce0da11e1b)
- Data as of 2026-08-30T08:39:29.467469+00:00.