@runbit/rag

Retrieval as a Runnable, built on the @runbit/memory interfaces.

npm install @runbit/rag @runbit/memory

import { knowledge } from "@runbit/rag";

const docs = knowledge({
  vectorStore,   // implements @runbit/memory VectorStore
  embeddings,    // { embed(text) => number[] }
  reranker,      // optional
  topK: 5,
});

const { matches } = await app.run(docs, { query: "How do refunds work?" });

Retrieves, optionally reranks, and emits memory:read. Compose it with an agent() to ground answers in your data.

Part of Runbit — the AI Runtime. Everything executes through one small, event-driven runtime; agents, workflows, RAG, and observability are packages on top.

• Umbrella: @runbit/runtime
• License: MIT