@augur-rag/core

The Augur SDK: adaptive retrieval orchestration for RAG and semantic search. Per-query routing across vector, BM25, hybrid, and cross-encoder reranking. A trace is returned with every search response.

Install

npm install @augur-rag/core
# also install the peer dep if you use LocalEmbedder or LocalReranker (on-device ONNX models):
npm install @huggingface/transformers

@huggingface/transformers is an optional peer dep. Only LocalEmbedder and LocalReranker need it. If you wire in OpenAI / Cohere / Voyage / Anthropic or any other provider via the Embedder / Reranker interface, you can skip it.

Hello world

import { Augur, LocalEmbedder } from "@augur-rag/core";

const augr = new Augur({ embedder: new LocalEmbedder() });

await augr.index([
  { id: "1", content: "Postgres supports vector search via pgvector." },
  { id: "2", content: "Pinecone is a managed vector database." },
]);

const { results, trace } = await augr.search({
  query: "How do I store vectors in Postgres?",
  topK: 5,
});

console.log(trace.decision.strategy);  // "vector"
console.log(trace.decision.reasons);   // ["natural-language question → semantic search", ...]

LocalEmbedder is an on-device sentence-transformer (Xenova/all-MiniLM-L6-v2 by default, ~22 MB ONNX, no API keys). Swap to a hosted provider by implementing the three-method Embedder interface.

Learn more

• Project README: pitch, BEIR comparison table, and quick start
• Architecture: how the orchestrator, router, adapters, chunkers, and rerankers fit together
• Examples: hosted embedders, contextual retrieval, pgvector setup, MMR diversity

License

MIT. See LICENSE.