https://github.com/wrannaman/rememberall

solving long term memory storage for llms
https://github.com/wrannaman/rememberall

llm long-term-memory memory

Last synced: 6 months ago
JSON representation

solving long term memory storage for llms

• Host: GitHub
• URL: https://github.com/wrannaman/rememberall
• Owner: wrannaman
• License: mit
• Created: 2024-12-11T00:23:38.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2025-01-06T16:29:37.000Z (9 months ago)
• Last Synced: 2025-03-30T03:06:35.840Z (6 months ago)
• Topics: llm, long-term-memory, memory
• Language: JavaScript
• Homepage: https://rememberall.xyz
• Size: 1.27 MB
• Stars: 120
• Watchers: 1
• Forks: 7
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# rememberall 🔮

A secure, open-source long-term memory solution for GPT Store builders and AI developers. Just mention `@rememberall` in your custom GPT to unlock persistent memory across conversations.

## For GPT Builders

1. Install from GPT Store

2. Add to your custom GPT's configuration using `./api/openapi.yaml`

3. Your GPT now has persistent memory!

### How It Works

#### 1. Store Memories

![Input Example](readme_assets/input.png)

*Your GPT can store important information from conversations*

#### 2. Load Past Context

![Load Memory](readme_assets/load_memory.png)

*Easily retrieve relevant memories when needed*

#### 3. Natural Interactions

![Remember Example](readme_assets/remember.png)

*Seamless integration with conversation flow*

#### 4. Manage Memories

![Dashboard UI](readme_assets/ui.png)

*Intuitive dashboard for memory management*

```typescript

// Example usage in your GPT's system prompt

When user mentions past conversations, use @rememberall to recall context:

User: "What did we discuss about authentication last week?"

Assistant: Let me check @rememberall

Assistant: According to our previous discussion, we implemented JWT-based auth...

```

## For Developers

### Features

- 🔒 **Privacy-First**: Self-host your memory store

- 🚀 **Vector-Based**: Efficient semantic search

- 🔗 **Easy Integration**: Simple REST API

- 📦 **Open Source**: Customize and extend as needed

### Quick Start

1. Deploy using Docker Compose:

```bash

git clone https://github.com/yourusername/rememberall.git

cd rememberall/deploy

docker-compose up -d

```

2. Available API Endpoints:

```http

# List or search memories

GET /memories?search=query&limit=10&offset=0

# Create new memory

POST /memory

{

    "memory": "Your memory text here"

}

```

3. Authentication:

- All endpoints require Bearer token authentication

- Include token in requests: `Authorization: Bearer your-jwt-token`

### API Response Examples

Get Memories:

```json

{

    "success": true,

    "memories": [

        {

            "id": "mem_123",

            "memory": "Discussion about authentication systems"

        }

    ]

}

```

Create Memory:

```json

{

    "success": true,

    "memory": {

        "id": "mem_124",

        "memory": "New project requirements discussion"

    }

}

```

## Architecture & Security

- End-to-end encryption

- Self-hosted vector storage

- Fine-grained access control

- GDPR-compliant data handling