AutoRAG

Why we included this project

AutoRAG 2.0 is a librarian agent for document collections rather than a conventional search tool. It runs an agent loop that searches and reads your PDFs, wikis, notes, and research papers, then reports back with clean, numbered knowledge units that carry page or section references, so you get an answer you can act on instead of a raw grep dump. Several retrieval strategies are available under one interface: lexical matching for config files, semantic vector search for dense prose, BM25 for legal and specification text, and a hybrid path for mixed collections, with everything merged into a single result set. A self-evolving memory system records which retrieval methods and document areas work for which kinds of queries, so the tool finds the right source faster the more you use it. It pays off most for teams with document-heavy workflows, whether analysts pulling findings from quarterly reports or researchers tracking a topic across dozens of papers.