wtf is rag-retrieval-tools?

This project is a large curated reference list covering retrieval-augmented generation, usually shortened to RAG, and the broader field of retrieval. Rather than being software you run, it catalogs over 200 tools that together make up a typical RAG stack: parsing and chunking source files, turning that content into searchable vectors or text indexes, retrieving relevant candidates, reranking them for quality, generating a grounded answer, and evaluating how well the whole system performed. The README is explicit that it separates these layers carefully, so a full RAG framework is listed apart from a vector database, an embedding model is listed apart from a reranker, and an evaluation toolkit is kept separate from an observability platform. Entries span sixteen categories, including frameworks for building end to end RAG applications, vector databases and search libraries, engines for lexical or hybrid keyword search, hosted managed retrieval services, embedding model providers, rerankers that improve result ordering, tools for parsing PDFs and scanned documents, web crawlers for extracting clean text, ingestion pipelines that keep an index in sync with source data, tools for retrieval over graphs of entities and relationships, multimodal retrieval across images and video, query optimization techniques, evaluation frameworks, fixed benchmarks for comparing systems, and tools for observability, security, and governance of a live RAG system. Every entry links back to its original source, such as a repository, documentation site, model card, or research paper, and each is labeled as open source, open weights, open core, or fully commercial. The catalog is also published as machine readable JSON and CSV files and is reviewed monthly, with a dedicated section listing tools that have been discontinued. The project is released under the CC0 license, meaning it sits in the public domain, and it accepts pull requests from contributors who want to add or correct entries.