Quickstart

Install vault-memory and store your first fact in under five minutes.

Get a single-agent Vault instance running with SQLite and OpenAI embeddings.

Install

npm install vault-memory drizzle-orm better-sqlite3

faiss-node installs automatically for vector search. drizzle-orm is a peer dependency — bring your own driver.

Create tables

For a quick test, create the claims table directly:

import Database from "better-sqlite3";
import { drizzle } from "drizzle-orm/better-sqlite3";
import { Memory } from "vault-memory";

const sqlite = new Database(":memory:");
sqlite.exec(`
  CREATE TABLE claims (
    id TEXT PRIMARY KEY,
    entity TEXT NOT NULL,
    text TEXT NOT NULL,
    type TEXT NOT NULL,
    source_agent TEXT NOT NULL,
    conversation_id TEXT NOT NULL,
    confidence REAL NOT NULL DEFAULT 1.0,
    status TEXT NOT NULL DEFAULT 'active',
    supersedes TEXT,
    embedding TEXT,
    created_at INTEGER NOT NULL,
    last_accessed_at INTEGER
  )
`);

const db = drizzle(sqlite);

For production, use Drizzle migrations from vault-memory/schema (claims, agent_authority, disputes, entities). See Schema.

Store and search

const memory = new Memory("agent-a", {
  db,
  apiKey: process.env.OPENAI_API_KEY,
});

const { added } = await memory.add(
  [{ role: "user", content: "I'm vegetarian" }],
  "conversation-1",
);
console.log(added);

const results = await memory.search("what does the user eat?");
console.log(results[0]?.text);

What happens under the hood

Extract — LLM pulls structured facts from the latest conversation turn.
Classify — Compare against similar claims from the same agent; write ADD, UPDATE, REPLACE, or skip (NOOP).
Index — Embed the fact and upsert into the FAISS vector store.

memory.search() is scoped to the constructing agent's sourceAgent. For cross-agent search, use the exported search() function — see Multi-agent setup.

Installation

Bun, Postgres, migrations, and vector store options.

How it works