Mem0

Add, search, list, and delete user memories across sessions with Mem0Tools using all-functions and restricted configurations.

Enable Agno agents to remember user preferences, past interactions, and specific facts across different conversations and platforms with Mem0.

  • Adaptive memory
  • User centric learning
  • Temporal context
  • Search and retrieval

Prerequisites

export MEM0_API_KEY="your_mem0_api_key_here"
export OPENAI_API_KEY="your_openai_api_key_here"
unset MEM0_ORG_ID MEM0_PROJECT_ID

from textwrap import dedent

from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.tools.mem0 import Mem0Tools

# ---------------------------------------------------------------------------
# Create Agent
# ---------------------------------------------------------------------------

USER_ID = "jane_doe"
SESSION_ID = "agno_session"

# Example 1: Enable all Mem0 functions
agent_all = Agent(
    model=OpenAIChat(id="gpt-4o"),
    tools=[
        Mem0Tools(
            all=True,  # Enable all Mem0 memory functions
        )
    ],
    user_id=USER_ID,
    session_id=SESSION_ID,
    markdown=True,
    instructions=dedent(
        """
        You have full access to memory operations. You can create, search, update, and delete memories.
        Proactively manage memories to provide the best user experience.
        """
    ),
)

# Example 2: Enable specific Mem0 functions only
agent_specific = Agent(
    model=OpenAIChat(id="gpt-4o"),
    tools=[
        Mem0Tools(
            enable_add_memory=True,
            enable_search_memory=True,
            enable_get_all_memories=False,
            enable_delete_all_memories=False,
        )
    ],
    user_id=USER_ID,
    session_id=SESSION_ID,
    markdown=True,
    instructions=dedent(
        """
        You can add new memories and search existing ones, but cannot delete or view all memories.
        Focus on learning and recalling information about the user.
        """
    ),
)

# Example 3: Default behavior with full memory access
agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    tools=[
        Mem0Tools(
            enable_add_memory=True,
            enable_search_memory=True,
            enable_get_all_memories=True,
            enable_delete_all_memories=True,
        )
    ],
    user_id=USER_ID,
    session_id=SESSION_ID,
    markdown=True,
    instructions=dedent(
        """
        You have an evolving memory of this user. Proactively capture new personal details,
        preferences, plans, and relevant context the user shares, and naturally bring them up
        in later conversation. Before answering questions about past details, recall from your memory
        to provide precise and personalized responses. Keep your memory concise: store only
        meaningful information that enhances long-term dialogue. If the user asks to start fresh,
        clear all remembered information and proceed anew.
        """
    ),
)

# Example usage with all functions enabled

# ---------------------------------------------------------------------------
# Run Agent
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    print("=== Example 1: Using all Mem0 functions ===")
    agent_all.print_response("I live in NYC and work as a software engineer")
    agent_all.print_response(
        "Summarize all my memories and delete outdated ones if needed"
    )

    # Example usage with specific functions only
    print("\n=== Example 2: Using specific Mem0 functions (add + search only) ===")
    agent_specific.print_response("I love Italian food, especially pasta")
    agent_specific.print_response("What do you remember about my food preferences?")

    # Example usage with default configuration
    print("\n=== Example 3: Default Mem0 agent usage ===")
    agent.print_response("I live in NYC")
    agent.print_response("I lived in San Francisco for 5 years previously")
    agent.print_response("I'm going to a Taylor Swift concert tomorrow")

    agent.print_response("Summarize all the details of the conversation")

    # More examples:
    # agent.print_response("NYC has a famous Brooklyn Bridge")
    # agent.print_response("Delete all my memories")
    # agent.print_response("I moved to LA")
    # agent.print_response("What is the name of the concert I am going to?")

Current Mem0 Platform compatibility

The current Platform client no longer accepts org_id or project_id. Leave them unset, including their environment variables. Agno's search_memory and get_all_memories still pass top-level user_id, which the current Platform API rejects. Until the adapter is updated, use explicit-filter client calls for reads:

from mem0 import MemoryClient

client = MemoryClient()
results = client.search(query="food preferences", filters={"user_id": USER_ID})
page = client.get_all(filters={"user_id": USER_ID}, page=1, page_size=50)
print(results)
print(page)

Follow the list response's pagination for a complete result. The source's immediate add-then-recall sequence also depends on the Platform finishing its asynchronous memory processing. Treat these as separate operations instead of running the unadapted full demo.

All three agents share USER_ID. SESSION_ID does not isolate Mem0 memories, and the toolkit's delete operation deletes every memory for the user, not selected outdated memories. Remove the cleanup prompt unless that is intended. No Agno conversation database/history is configured.

Run the Example

# Clone and setup repo
git clone https://github.com/agno-agi/agno.git
cd agno
git checkout d703c34f3abf3c41275d3fb2da6e0518a8881f24

# Create and activate virtual environment
./scripts/demo_setup.sh
source .venvs/demo/bin/activate
uv pip install -U mem0ai

python cookbook/91_tools/mem0_tools.py

For details, see Mem0 tools cookbook.