barq-web-rag

A document question-answering app that runs entirely in the browser. It searches and answers from your files locally, so no data leaves your machine.

Share on XLicense: Apache-2.0

Overview

Barq RAG is a document question-answering app that runs entirely in the browser. It uses Rust compiled to WebAssembly for the retrieval engine, vector storage and compute kernels, and runs local language models in the browser through transformers.js and WebGPU. Documents are processed, embedded and answered on your device, with no external API and no data leaving your machine.

Key features

  • Hybrid search combining BM25 and semantic search with RRF reranking
  • Ingests PDF, DOCX, Markdown and TXT files
  • Local vector index with persistence in the browser
  • Embeddings and LLM inference run on your device

Best for

People who want to ask questions about private documents without sending them to a server. Setup is npm install and a dev server.

Upstream
YASSERRMD/barq-web-rag
Fork on GitHub
Guo-astro/barq-web-rag
Upstream stars
29
Category
AI agents and LLM tools
Language
JavaScript
License
Apache-2.0
Forked
2026-03-11
Sync status
In syncLast synced 2026-09-29