@open-engram/core

Core memory architecture for Open-Engram — a biologically-grounded, four-store memory system for AI agents.

Install

npm install @open-engram/core

Overview

Implements the four-store memory model inspired by Standard Consolidation Theory, Complementary Learning Systems, Baddeley's working memory model, and thalamic gating theory:

• Sensory Buffer — high-throughput ingestion with TTL-based expiry
• Working Memory — capacity-limited active context scored by Recency-Frequency-Relevance (RFR)
• Episodic Store — contextually-indexed event records with vector search
• Semantic Store — long-term structured knowledge with versioning and conflict resolution

Plus subsystems: Attention Gate, Consolidation Engine (7-stage distillation pipeline), and Tiered Retrieval (T0/T1/T2).

Quick Start

import { EngramClient } from '@open-engram/core';

const client = await EngramClient.create({
  workingMemory: { tokenBudget: 8192 },
  consolidation: { trigger: 'session-end' },
});

// Ingest an event
await client.sense({ content: 'User prefers dark mode', source: 'settings' });

// Run attention gate → promote to working memory
await client.focus();

// Search across stores
const results = await client.recall('user preferences', { limit: 5 });

// Save a fact to semantic memory
await client.remember('User prefers dark mode', { confidence: 0.9 });

// Run consolidation pipeline (episodic → semantic)
await client.consolidate();

// Export full memory state
const snapshot = await client.export();

API

| Method | Description | |---|---| | sense(event) | Ingest into sensory buffer | | focus() | Run attention gate, promote to working memory | | recall(query, opts) | Search across episodic + semantic stores | | remember(content, opts) | Write directly to semantic store | | checkpoint(label) | Snapshot working memory to episodic store | | consolidate() | Run 7-stage distillation pipeline | | forget(id) | Remove a specific record | | purge() | Clear all stores | | pin(id) / unpin(id) | Protect entries from eviction | | status() | Memory stats and store counts | | export() / import(data) | Serialize/deserialize full state |

Adapters

The core package uses an in-memory adapter by default. For persistent storage, use:

• @open-engram/adapters-storage — SQLite + sqlite-vec
• @open-engram/adapters-storage-postgres — PostgreSQL + pgvector
• @open-engram/adapters-storage-powersync — PowerSync bi-directional sync
• @open-engram/adapters-storage-supabase — Supabase + pgvector RPC

License

Apache-2.0