Retrieval-Augmented Generation (RAG) Workflows in R with Local and Web Search

Enables Retrieval-Augmented Generation (RAG) workflows in R by combining local vector search using 'DuckDB' with optional web search via the 'Tavily' API. Supports 'OpenAI'- and 'Ollama'-compatible embedding models, full-text and 'HNSW' (Hierarchical Navigable Small World) indexing, and modular large language model (LLM) invocation. Designed for advanced question-answering, chat-based applications, and production-ready AI pipelines. This package is the R equivalent of the 'python' package 'RAGFlowChain' available at < https://pypi.org/project/RAGFlowChain/>.

RAGFlowChainR

Overview

RAGFlowChainR is an R package for Retrieval-Augmented Generation (RAG) workflows with local retrieval backends (DuckDB and VectrixDB) plus optional web search.

The README is intentionally short. Full backend workflows are documented in vignettes.

Installation

install.packages("RAGFlowChainR")

Development version

install.packages("remotes")
remotes::install_github("knowusuboaky/RAGFlowChainR")

Backend Guides

DuckDB backend article: https://knowusuboaky.github.io/RAGFlowChainR/articles/duckdb-backend.html
VectrixDB backend article: https://knowusuboaky.github.io/RAGFlowChainR/articles/vectrixdb-backend.html
Function reference: https://knowusuboaky.github.io/RAGFlowChainR/reference/

Quick Start

library(RAGFlowChainR)

rag <- create_rag_chain(
  llm = function(prompt) "mock answer",
  vector_database_directory = "my_vectors.duckdb",
  method = "DuckDB",
  use_web_search = FALSE
)

rag$invoke("What is RAG?")
rag$disconnect()

For complete ingestion, indexing, and backend-specific setup, use the two backend vignettes above.

Environment Setup

Sys.setenv(TAVILY_API_KEY    = "your-tavily-api-key")
Sys.setenv(OPENAI_API_KEY    = "your-openai-api-key")
Sys.setenv(GROQ_API_KEY      = "your-groq-api-key")
Sys.setenv(ANTHROPIC_API_KEY = "your-anthropic-api-key")