memorymachines 0.2.0

Official Python SDK for Memory Machines API

Memory Machines Python SDK

Official Python SDK for the Memory Machines API.

Installation

pip install memorymachines

Or install from source:

git clone https://github.com/memorymachines/memorymachines-sdk.git
cd memorymachines-sdk
pip install -e .

Quick Start

# One-time login (saves to ~/.memorymachines/config.yaml)
memorymachines login

from memorymachines import MemoryClient

# No API key needed - auto-discovered from config file
client = MemoryClient()

# Ingest content
client.ingest("Had a great meeting about Q4 goals today")

# Search memories
memories = client.search("what did we discuss about Q4?")
for memory in memories:
    print(memory['content'])

Getting an API Key

1. Sign up at memorymachines.com
2. Go to your dashboard
3. Copy your API key (starts with mm_sk_)

Authentication

The SDK auto-discovers your API key in this order:

1. Explicit api_key parameter
2. MM_API_KEY environment variable
3. ~/.memorymachines/config.yaml file

CLI Login (Recommended)

# Save your API key securely
memorymachines login

# Check who's logged in
memorymachines whoami

# List profiles
memorymachines profiles

# Logout
memorymachines logout

Programmatic Login

from memorymachines import login, logout

# Save token
login("mm_sk_...")

# Remove token
logout()

Multiple Profiles

# Save multiple accounts
memorymachines login --profile personal
memorymachines login --profile work

# Switch profiles
memorymachines use work

# Use specific profile
client = MemoryClient(profile="work")

Usage

Initialize Client

from memorymachines import MemoryClient

# Auto-discover API key (recommended)
client = MemoryClient()

# Or explicit API key
client = MemoryClient(api_key="mm_sk_...")

# Or specific profile
client = MemoryClient(profile="work")

# Custom endpoint (for development)
client = MemoryClient(base_url="http://localhost:8080")

Ingest Data

# Single item
client.ingest(
    content="Meeting notes from today...",
    source="notes",
    metadata={"date": "2024-10-24"}
)

# Batch ingestion
items = [
    {'content': 'Email from client...', 'source': 'email'},
    {'content': 'Meeting notes...', 'source': 'notes'}
]
client.ingest_batch(items)

Search Memories

# Natural language search
memories = client.search("what did I discuss last week?", limit=10)

# List all memories with filters
response = client.list_memories(
    limit=50,
    source="email",
    since=datetime(2024, 10, 1)
)

Manage Memories

# Get specific memory
memory = client.get_memory(memory_id="mem_123")

# Delete memory
client.delete_memory(memory_id="mem_123")

# Check connected services
services = client.get_connected_services()  # ['gmail', 'slack']

Examples

See the examples/ directory for more usage examples:

• basic_usage.py - Basic SDK operations
• batch_ingest.py - Batch data ingestion

API Reference

MemoryClient

__init__(api_key: str, base_url: str = "https://api.memorymachines.com")

Initialize the client.

ingest(content: str, source: str = "custom", metadata: dict = None) -> dict

Ingest text content to extract memories.

ingest_batch(items: List[dict]) -> dict

Ingest multiple items at once.

search(query: str, limit: int = 10) -> List[dict]

Search memories with natural language.

list_memories(limit: int = 50, offset: int = 0, source: str = None, since: datetime = None) -> dict

List memories with filters.

get_memory(memory_id: str) -> dict

Get a specific memory by ID.

delete_memory(memory_id: str) -> dict

Delete a memory.

get_connected_services() -> List[str]

Get list of connected OAuth services.

health_check() -> bool

Check if API is accessible.

Development

Setup

git clone https://github.com/memorymachines/memorymachines-sdk.git
cd memorymachines-sdk
pip install -e ".[dev]"

Run Examples

export MEMORY_MACHINES_API_KEY="mm_sk_..."
python examples/basic_usage.py

Support

• Documentation: docs.memorymachines.com
• Issues: GitHub Issues
• Email: [e-mail]

License

Apache 2.0 - see LICENSE file for details.