[BUG] AttributeError in CrewAI MemoryClient using Gemini Api Key with Mem0 (self.um.search())

[balochan970]

Description

I encountered an issue when using CrewAI with Mem0 as the memory provider and Google Gemini 2.0 as the LLM The error occurs in ContextualMemory when trying to fetch user memory.

I tried the code with

client = MemoryClient()

and

client = MemoryClient(
api_key="api key directly",
org_id="my org id",
project_id="my proj id"
)

Steps to Reproduce

Install crewai and mem0ai using: pip install crewai mem0ai

import os
from crewai import Crew, Process
from mem0 import MemoryClient
os.environ["MEM0_API_KEY"] = userdata.get('MEM0_API_KEY')
client = MemoryClient()
messages = [
{"role": "user", "content": "Hi there! I'm planning a vacation and could use some advice."},
{"role": "assistant", "content": "Hello! I'd be happy to help with your vacation planning. What kind of destination do you prefer?"},
{"role": "user", "content": "I am more of a beach person than a mountain person."},
{"role": "assistant", "content": "That's interesting. Do you like hotels or Airbnb?"},
{"role": "user", "content": "I like Airbnb more."},
]
client.add(messages, user_id="john")

agent = Agent(
role="About User",
goal="You know everything about the user.",
backstory="""You are a master at understanding people and their preferences.""",
verbose=True,
allow_delegation=False,
llm=llm1, #the llm1 is saved in the os.environ
)
task = Task(
description="Answer the following questions about the user: {question}",
expected_output="An answer to the question.",
agent=agent,
)

crew = Crew(
agents=[agent],
tasks=[task],
process = Process.sequential,
verbose=True,
memory=True,
memory_config={
"provider": "mem0",
"config": {"user_id": "john"},
},
)

crew.kickoff(inputs={"question": "What is your favorite vacation destination?"})

Expected behavior

Expected Behavior:
The CrewAI agent should retrieve user memory from Mem0.

Actual Behavior:
AttributeError: 'NoneType' object has no attribute 'search'

Screenshots/Code snippets

Memory is not properly fetched, and self.um.search() raises an AttributeError.

Operating System

Windows 11

Python Version

3.12

crewAI Version

0.108.0

crewAI Tools Version

0.36.0

Virtual Environment

Venv

Evidence

╭──────────────────────────────────────────── Crew Execution Started ─────────────────────────────────────────────╮
│ │
│ Crew Execution Started │
│ Name: crew │
│ ID: 8e2eaa6a-2452-4dd1-9c32-44202e69dc7d │
│ │
│ │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
🚀 Crew: crew
└── 📋 Task: bfa624c9-c5a8-49cd-adbd-4ebf815c197f
Status: Executing Task...
🚀 Crew: crew
└── 📋 Task: bfa624c9-c5a8-49cd-adbd-4ebf815c197f
Assigned to: About User
Status: ❌ Failed
╭───────────────────────────────────────────────── Task Failure ──────────────────────────────────────────────────╮
│ │
│ Task Failed │
│ Name: bfa624c9-c5a8-49cd-adbd-4ebf815c197f │
│ Agent: About User │
│ │
│ │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭───────────────────────────────────────────────── Crew Failure ──────────────────────────────────────────────────╮
│ │
│ Crew Execution Failed │
│ Name: crew │
│ ID: 8e2eaa6a-2452-4dd1-9c32-44202e69dc7d │
│ │
│ │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

AttributeError Traceback (most recent call last)
in <cell line: 0>()
47 )
48
---> 49 result = crew.kickoff(inputs={"question": "What is your favorite vacation destination?"})
50
51 # ✅ Output print karna

8 frames
/usr/local/lib/python3.11/dist-packages/crewai/memory/contextual/contextual_memory.py in _fetch_user_context(self, query)
95 str: Formatted user memories as bullet points, or an empty string if none found.
96 """
---> 97 user_memories = self.um.search(query)
98 if not user_memories:
99 return ""

AttributeError: 'NoneType' object has no attribute 'search'

Possible Solution

class ContextualMemory:
def init(
self,
memory_config: Optional[Dict[str, Any]],
stm: ShortTermMemory,
ltm: LongTermMemory,
em: EntityMemory,
um: UserMemory,
):
if memory_config is not None:
self.memory_provider = memory_config.get("provider")
# Testing the Mem0 client as if provider is "mem0"
if self.memory_provider == "mem0": #<-----Here, but this is a manual workaround
self.um = memory_config["config"].get("client")

            self.search_kwargs = memory_config["config"].get("search_kwargs", {})
        else:
            self.um = um
    else:
        self.memory_provider = None
        self.um = um

Additional context

The issue might be due to self.um being None in ContextualMemory.

[Vidit-Ostwal]

Hi @balochan970
For a quick fix, can you try something like this

crew = Crew(
agents=[agent],
tasks=[task],
process = Process.sequential,
verbose=True,
memory=True,
memory_config={
"provider": "mem0",
"config": {"user_id": "john", api_key = os.environ["MEM0_API_KEY"]},
"user_memory" : {}
}

I will try to update the docs as necessary.
Let me know if this works

[balochan970]

Thanks, @Vidit-Ostwal !
Your fix worked perfectly. The issue was indeed with how CrewAI handles mem0 API key initialization. Explicitly passing it inside memory_config solved the problem. It'd be great if the docs could clarify this behavior to help others avoid similar confusion. Appreciate the quick response!"

P.S. Here's the corrected syntax for reference.

crew = Crew(
agents=[agent],
tasks=[task],
process=Process.sequential,
verbose=True,
memory=True,
memory_config={
"provider": "mem0",
"config": {
"user_id": "john",
"api_key": os.environ["MEM0_API_KEY"]
},
"user_memory": {}
}
)

[Vidit-Ostwal]

Hi @balochan970, thanks for this,
I have raised this PR which I believe will get this sorted, will try to add the documentation along as well.

I have added some way to use user_memory in the PR comment section, can you check it once, and share whether do you think this way of passing argument's is good or not from a user perspective.

Thanks a lot!