RAGxplorer is an open-source tool for building interactive visualisations of Retrieval Augmented Generation (RAG) systems. Created by Gabriel Chua and hosted on GitHub with about 1,200 stars, it helps developers literally see how their documents are chunked, embedded and retrieved, instead of debugging a RAG pipeline blind. The project was last updated in early 2025.
When a RAG application gives wrong answers, the cause is often invisible: chunks may be split awkwardly, embeddings may cluster poorly, or the retriever may surface irrelevant passages. RAGxplorer attacks that problem visually. You load your documents, configure chunking and embedding settings, and the Streamlit app projects the resulting vectors into an interactive two-dimensional map you can explore. You can inspect individual chunks, run test queries and see which passages are retrieved and how close they sit to the query in embedding space. That makes it much easier to tune chunk sizes, choose embedding models and spot gaps or duplication in a knowledge base before shipping. It is a development and diagnostics companion rather than a production search engine.
Because it runs locally as a Streamlit app, RAGxplorer fits naturally at the start of a RAG project: explore a corpus, settle on chunking and embedding choices with visual feedback, then carry those settings into your production indexing pipeline with far more confidence.
RAGxplorer is completely free and open source under the MIT license. There is no hosted version or paid plan; you run it locally, though embedding or LLM API calls you configure may incur charges from those providers.
RAG developers, data scientists and AI engineers who want to understand and tune how their documents are indexed and retrieved before deploying a question-answering or search product.