
Vectorless RAG engine that finds answers by reasoning over a document's page tree
- Open source
- Open-source agent
- 40.5K GitHub stars
- Files and documents

Vectorless RAG engine that finds answers by reasoning over a document's page tree

Open-source knowledge framework for RAG Q&A, agent tasks and auto-built wikis

Lets your agent ingest, index and query document collections with the NeMo Retriever CLI
Local-first RAG that answers from your documents with page and section citations
Runs locallySearch a local folder of documents and check answers against them, with no API key
Runs locallySplit Markdown into heading-aware chunks for RAG, with paths and line numbers
Runs locallyGives agents a PDF's heading tree and section text so they read only what they need
Runs locallyConvert documents, images and web pages to Markdown from any MCP client
Runs locallyLocal MCP server that lets AI agents read PDFs and specs by section with cited quotes
Runs locally
Local AI search over your files, code and crawled sites, with cited answers and a built
Runs locally