@namitjain.india/agent-memory-sqlite

SQLite adapter for @namitjain.india/agent-memory. Provides persistent storage using better-sqlite3 with full-text search capabilities.

GitHub | Report Bug | Request Feature

Installation

npm install @namitjain.india/agent-memory-sqlite better-sqlite3 sqlite-vss

Usage

import { AgentMemory } from "@namitjain.india/agent-memory";
import { SQLiteAdapter } from "@namitjain.india/agent-memory-sqlite";

const memory = new AgentMemory({
  adapter: new SQLiteAdapter({
    dbPath: "./memory.db"
  }),
  embedding: {
    embedFn: async (text) => [text.length / 100, 0.5]
  }
});

SQLiteAdapter Options

| Option | Type | Default | Description | |--------|------|---------|-------------| | dbPath | string | :memory: | Path to SQLite database file | | database | SqliteDatabaseLike | - | Custom database instance | | vssExtensionPath | string | - | Path to sqlite-vss extension | | loadVss | boolean | false | Load VSS extension for vector search |

Features

• Persistent Storage: Data survives process restarts
• Session-Based: Automatic session indexing
• Type Support: Full support for entries, facts, and summaries
• VSS Support: Optional vector similarity search via sqlite-vss
• Schema Auto-Creation: Tables and indexes created automatically

Database Schema

CREATE TABLE memory_items (
  id TEXT PRIMARY KEY,
  kind TEXT NOT NULL,
  session_id TEXT NOT NULL,
  timestamp INTEGER NOT NULL,
  importance REAL NOT NULL DEFAULT 0.5,
  role TEXT,
  content TEXT,
  key_name TEXT,
  value_text TEXT,
  embedding TEXT,
  metadata TEXT,
  from_timestamp INTEGER,
  to_timestamp INTEGER,
  replaced_entry_ids TEXT
);

CREATE INDEX idx_memory_items_session ON memory_items(session_id, timestamp);
CREATE INDEX idx_memory_items_kind ON memory_items(kind);

Example with VSS

import { AgentMemory } from "@namitjain.india/agent-memory";
import { SQLiteAdapter } from "@namitjain.india/agent-memory-sqlite";

// Load sqlite-vss extension for vector similarity search
const memory = new AgentMemory({
  adapter: new SQLiteAdapter({
    dbPath: "./memory.db",
    vssExtensionPath: "/path/to/sqlite-vss.so",
    loadVss: true
  }),
  embedding: {
    embedFn: async (text) => [/* embeddings */]
  }
});

Methods

close()

Close the database connection.

const adapter = new SQLiteAdapter({ dbPath: "./memory.db" });
// ... use adapter ...
adapter.close();

Requirements

• Node.js >= 18.0.0
• better-sqlite3 >= 11.0.0
• sqlite-vss >= 0.1.2 (optional, for vector search)

Contributing & Collaboration

We welcome contributions, feedback, and feature requests!

• Bug Reports: Open an issue
• Feature Requests: Share it
• Pull Requests: Submit a PR

If this project helps you, please consider starring it on GitHub!

License

MIT