memorylayer-py 0.1.1

Privacy-first memory API for LLMs

rec0 Python SDK

Official Python client library for rec0 - Persistent memory infrastructure for AI applications.

Quick Start

from rec0 import Memory

# Initialize with your API key
mem = Memory(
    api_key='r0_live_sk_...',
    user_id='user_123'
)

# Store a memory
mem.store("User prefers dark mode and Italian food")

# Recall relevant memories
context = mem.recall(query="user preferences")
print(context)
# Returns: "User prefers dark mode and Italian food"

Install

pip install memorylayer-py

Note: The package name is memorylayer-py but you import it as rec0:

from rec0 import Memory  # Import name is 'rec0', not 'memorylayer'

Get Your API Key

1. Sign up at rec0.vercel.app
2. Get your API key from the dashboard
3. Start building!

Documentation

• Quick Start Guide
• API Reference
• Examples
• Full Documentation

Features

• Auto-Capture: Automatically extracts entities, facts, and preferences
• Smart Retrieval: Hybrid BM25 + cosine similarity search
• Memory Decay: Intelligent aging and conflict resolution
• Privacy-first: Per-user encryption, data isolation, and GDPR-compliant deletion. Your users' data is encrypted and never shared with third parties.
• LLM-Agnostic: Works with any LLM (OpenAI, Anthropic, Google, local models)

rec0 vs Mem0

rec0	Mem0
Privacy: Per-user encryption & isolation	Standard cloud storage

Enterprise Deployment

The Python SDK is fully open source (MIT license). The backend API is closed source but available for enterprise on-premise deployment.

For self-hosted deployments, contact: [e-mail]

Examples

Basic Usage

from rec0 import Memory

mem = Memory(api_key='r0_...', user_id='user_123')

# Store
mem.store("User is a Python developer")

# Recall
context = mem.recall(query="programming")

AI Chatbot with Memory

from rec0 import Memory
from openai import OpenAI

mem = Memory(api_key='r0_...', user_id='user_456')
ai = OpenAI()

# Get relevant context
context = mem.recall(query=user_message)

# Send to LLM with context
response = ai.chat.completions.create(
    model="gpt-4",
    messages=[
        {"role": "system", "content": f"Context: {context}"},
        {"role": "user", "content": user_message}
    ]
)

# Store the conversation
mem.store(f"User: {user_message}\nAssistant: {response}")

See examples/ for more.

API Reference

Memory(api_key, user_id, app_id=None, base_url=None)

Initialize the memory client.

Parameters:

• api_key (str): Your rec0 API key
• user_id (str): Unique identifier for the user
• app_id (str, optional): Application identifier
• base_url (str, optional): Custom API endpoint

store(content, metadata=None)

Store a memory.

Parameters:

• content (str): The content to remember
• metadata (dict, optional): Additional metadata

Returns: dict with memory_id

recall(query, limit=10)

Retrieve relevant memories.

Parameters:

• query (str): Search query
• limit (int): Maximum number of results

Returns: list of memory objects

delete(memory_id=None, user_id=None)

Delete memories.

Parameters:

• memory_id (str, optional): Specific memory to delete
• user_id (str, optional): Delete all memories for user

Contributing

Contributions are welcome! See CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE file for details.

Links

• rec0.vercel.app
• GitHub
• Docs

Support

• GitHub Issues: Report bugs
• Email: [e-mail]

---

Made with ❤️ by rec0