The Memory Layer for AI Agents - Drop-in memory infrastructure for AI agents and apps. Context that persists. Built for production.
Python agentic-memoryagentic-memory-systemagentsaiai-agentschatgptgenaillmlong-term-memorymemorymemory-managementpythonragstate-management
import builtins
import importlib
import inspect
import sys
from unittest.mock import Mock, patch
import pytest
from mem0 import Memory, MemoryClient
from mem0.proxy.main import Chat, Completions, Mem0
@pytest.fixture
def mock_memory_client():
mock_client = Mock(spec=MemoryClient)
mock_client.user_email = None
return mock_client
@pytest.fixture
def mock_openai_embedding_client():
with patch("mem0.embeddings.openai.OpenAI") as mock_openai:
mock_client = Mock()
mock_openai.return_value = mock_client
yield mock_client
@pytest.fixture
def mock_openai_llm_client():
with patch("mem0.llms.openai.OpenAI") as mock_openai:
mock_client = Mock()
mock_openai.return_value = mock_client
yield mock_client
@pytest.fixture
def mock_litellm():
with patch("mem0.proxy.main.litellm") as mock:
yield mock
def test_mem0_initialization_with_api_key(mock_openai_embedding_client, mock_openai_llm_client):
mem0 = Mem0()
assert isinstance(mem0.mem0_client, Memory)
assert isinstance(mem0.chat, Chat)
def test_mem0_initialization_with_config():
config = {"some_config": "value"}
with patch("mem0.Memory.from_config") as mock_from_config:
mem0 = Mem0(config=config)
mock_from_config.assert_called_once_with(config)
assert isinstance(mem0.chat, Chat)
def test_mem0_initialization_without_params(mock_openai_embedding_client, mock_openai_llm_client):
mem0 = Mem0()
assert isinstance(mem0.mem0_client, Memory)
assert isinstance(mem0.chat, Chat)
def test_chat_initialization(mock_memory_client):
chat = Chat(mock_memory_client)
assert isinstance(chat.completions, Completions)
def test_completions_create(mock_memory_client, mock_litellm):
completions = Completions(mock_memory_client)
messages = [{"role": "user", "content": "Hello, how are you?"}]
mock_memory_client.search.return_value = [{"memory": "Some relevant memory"}]
mock_litellm.completion.return_value = {"choices": [{"message": {"content": "I'm doing well, thank you!"}}]}
mock_litellm.supports_function_calling.return_value = True
response = completions.create(model="gpt-4.1-nano-2025-04-14", messages=messages, user_id="test_user", temperature=0.7)
mock_memory_client.add.assert_called_once()
mock_memory_client.search.assert_called_once()
mock_litellm.completion.assert_called_once()
call_args = mock_litellm.completion.call_args[1]
assert call_args["model"] == "gpt-4.1-nano-2025-04-14"
assert len(call_args["messages"]) == 2
assert call_args["temperature"] == 0.7
assert response == {"choices": [{"message": {"content": "I'm doing well, thank you!"}}]}
def test_completions_create_keeps_filters_out_of_add(mock_memory_client, mock_litellm):
completions = Completions(mock_memory_client)
mock_memory_client.search.return_value = []
mock_litellm.supports_function_calling.return_value = True
with patch("mem0.proxy.main.threading.Thread") as mock_thread:
mock_thread.side_effect = lambda target, daemon: Mock(start=target)
completions.create(
model="gpt-4.1-nano-2025-04-14",
messages=[{"role": "user", "content": "Hello"}],
user_id="test_user",
filters={"categories": {"in": ["travel"]}},
)
assert "filters" not in mock_memory_client.add.call_args.kwargs
assert mock_memory_client.add.call_args.kwargs["user_id"] == "test_user"
assert mock_memory_client.search.call_args.kwargs["filters"] == {"categories": {"in": ["travel"]}}
def test_completions_create_with_system_message(mock_memory_client, mock_litellm):
completions = Completions(mock_memory_client)
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Hello, how are you?"},
]
mock_memory_client.search.return_value = [{"memory": "Some relevant memory"}]
mock_litellm.completion.return_value = {"choices": [{"message": {"content": "I'm doing well, thank you!"}}]}
mock_litellm.supports_function_calling.return_value = True
completions.create(model="gpt-4.1-nano-2025-04-14", messages=messages, user_id="test_user")
call_args = mock_litellm.completion.call_args[1]
assert call_args["messages"][0]["role"] == "system"
assert call_args["messages"][0]["content"] == "You are a helpful assistant."
def test_completions_create_messages_default_does_not_leak_between_calls(mock_memory_client, mock_litellm):
"""Regression test for the B006 mutable-default bug in Completions.create.
Before the fix, `messages: List = []` made every call that didn't pass
`messages` share the same module-level list. A previous call could mutate
that list (e.g. via `_prepare_messages`) and subsequent calls would observe
the leaked state instead of an empty list.
After the fix, `messages` defaults to `None` and is normalized to a fresh
`[]` inside the function on each call, isolating call N from call N-1.
"""
completions = Completions(mock_memory_client)
mock_litellm.supports_function_calling.return_value = True
mock_litellm.completion.return_value = {"choices": [{"message": {"content": "ok"}}]}
mock_memory_client.search.return_value = []
# Each call passes a fresh list — confirms the public happy path stays green.
completions.create(
model="gpt-4.1-nano-2025-04-14",
messages=[{"role": "user", "content": "first"}],
user_id="user_a",
)
completions.create(
model="gpt-4.1-nano-2025-04-14",
messages=[{"role": "user", "content": "second"}],
user_id="user_b",
)
# The Completions.create signature must not bind a mutable container as
# the default for `messages`. This is what B006 lints against and what the
# historical default `messages: List = []` violated.
sig = inspect.signature(Completions.create)
messages_default = sig.parameters["messages"].default
assert messages_default is None, (
f"Completions.create(messages=...) must default to None to avoid the "
f"B006 shared-default-list bug; got {messages_default!r}."
)
def test_missing_litellm_raises_actionable_import_error(monkeypatch):
monkeypatch.delitem(sys.modules, "mem0.proxy.main", raising=False)
monkeypatch.delitem(sys.modules, "litellm", raising=False)
real_import = builtins.__import__
def fake_import(name, *args, **kwargs):
if name == "litellm":
raise ImportError("No module named 'litellm'")
return real_import(name, *args, **kwargs)
monkeypatch.setattr(builtins, "__import__", fake_import)
with pytest.raises(ImportError, match="pip install litellm"):
importlib.import_module("mem0.proxy.main")