Installation

Install vault-memory with SQLite or Postgres, migrations, and vector backends.

Requirements

Node.js 18+ or Bun 1.2+
OpenAI API key (or custom embed / llm functions)
Drizzle-compatible database driver

npm

npm install vault-memory drizzle-orm better-sqlite3

Bun

If Bun blocks the faiss-node native postinstall:

bun pm trust faiss-node
bun install

Schema and migrations

Import tables from the package schema:

import { claims, agentAuthority, disputes } from "vault-memory/schema";

Generate migrations with Drizzle Kit:

npx drizzle-kit generate
npx drizzle-kit push

Table	Purpose
`claims`	Facts with status, embeddings, attribution
`agent_authority`	Per-entity trust weights for multi-agent resolution
`disputes`	Audit trail when authority cannot pick a winner
`entities`	Canonical entity taxonomy (optional, for `useEntityRegistry`)

Full column reference: Schema.

Vector store

Default: FaissVectorStore via faiss-node — local index, SQL rows as source of truth.

Opt out (e.g. CI without native bindings):

import { Memory, MemoryVectorStore } from "vault-memory";

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

Postgres: use PgVectorStore from vault-memory/pg when you want the index in the database.

See Vector store for backends, multi-process rebuild, and index lifecycle.

Environment variables

OPENAI_API_KEY=sk-...

Override models via MemoryConfig:

const memory = new Memory("agent-a", {
  db,
  apiKey: process.env.OPENAI_API_KEY,
  model: "gpt-4o-mini",
  embeddingModel: "text-embedding-3-small",
});