Quick Reference

Fresh

SDK Installation

Python

pip install mem0ai

Node.js

npm install mem0ai

Platform Client

Python

from mem0 import MemoryClient
client = MemoryClient(api_key="your-api-key")

JavaScript

import MemoryClient from 'mem0ai';
const client = new MemoryClient({ apiKey: 'your-api-key' });

OSS Client

Python

from mem0 import Memory
m = Memory()

JavaScript

import { Memory } from 'mem0ai/oss';
const memory = new Memory();

Core Operations Cheat Sheet

Add

client.add(messages, user_id="alice")
client.add(messages, user_id="alice", enable_graph=True)
client.add(messages, user_id="alice", metadata={"category": "travel"})

Search

client.search("query", filters={"user_id": "alice"})
client.search("query", filters={"AND": [
    {"user_id": "alice"},
    {"categories": {"contains": "travel"}}
]})

Update

client.update(memory_id="id", text="new content")
client.batch_update([{"memory_id": "id1", "text": "new1"}, ...])

Delete

client.delete(memory_id="id")
client.delete_all(user_id="alice")
client.batch_delete([{"memory_id": "id1"}, ...])

Get

client.get_all(filters={"user_id": "alice"})
client.get(memory_id="id")

Filter Syntax (Platform v2)

# Single filter
filters = {"user_id": "alice"}

# AND logic
filters = {"AND": [
    {"user_id": "alice"},
    {"agent_id": "chatbot"}
]}

# Date range
filters = {"AND": [
    {"user_id": "alice"},
    {"created_at": {"gte": "2024-07-01"}}
]}

# Categories
filters = {"AND": [
    {"user_id": "alice"},
    {"categories": {"contains": "travel"}}
]}

Environment Variables

# Platform
export MEM0_API_KEY="your-api-key"

# OSS (OpenAI default)
export OPENAI_API_KEY="your-openai-key"

# OSS (other providers)
export ANTHROPIC_API_KEY="your-key"
export COHERE_API_KEY="your-key"
export QDRANT_API_KEY="your-key"

Common Troubleshooting

Problem	Solution
Empty search results	Check embedder model matches vector store dimensions
Auth errors	Verify API key format (m0- prefix for platform)
Import errors	Run pip install --upgrade mem0ai
Duplicate memories	Use infer=True (default) for deduplication
Slow searches	Enable reranker, limit top_k to 10-20
Graph not working	Set enable_graph=True or enable at project level