Setup & Installation

This guide will help you set up and run the Memory engine locally.

Prerequisites

Before starting, ensure you have the following installed:

• Bun: The JavaScript runtime used by this project.
• PostgreSQL: Rational database for storing memory metadata.
• Qdrant: Vector database for storing embeddings.

Installation

1. Clone the repository:

   git clone https://github.com/sincerelyyyash/memory
   cd memory

2. Install dependencies:

   bun install

Configuration

1. Copy the example environment file:

   cp .env.example .env

2. Configure the environment variables in .env:

   Variable	Description	Default
   PORT	API Port	8000
   OPENROUTER_API_KEY	API Key for LLM services	-
   DATABASE_URL	PostgreSQL connection string	-
   QDRANT_URL	URL of your Qdrant instance	-
   QDRANT_API_KEY	API Key for Qdrant (if auth enabled)	-
   EMBEDDING_MODEL	Model for generating embeddings	openai/text-embedding-3-small
   EMBEDDING_DIMENSION	Dimension of the embedding vectors	Matches model
   ANSWER_MODEL	Model for RAG answer generation	gpt-4o-mini
   FACT_MODEL	Model for fact extraction	gpt-4o-mini
   QDRANT_COLLECTION_NAME	Name of the Qdrant collection	-
   RERANK_MODEL	Model for reranking	gpt-4o-mini
   RERANK_ENABLED	Enable reranking by default	false
   RERANK_TOP_K	Number of documents to rerank	-
   PROCEDURAL_MODEL	Model for procedural summary generation	gpt-4o-mini
   NODE_ENV	Environment (e.g., development, production)	-

Database Setup

1. Generate the Prisma client:

   bun x prisma generate

2. Push the schema to your database:

   bun x prisma db push

Running the Engine

Start the development server:

bun run index.ts

The server will start on the configured PORT (default: 8000).