mem0-mcp-server
mem0-mcp-server — exposes Mem0 persistent semantic memory as an MCP HTTP server; supports add/search/read/update/delete operations and semantic search for agent memory.
mem0-mcp-server
MCP server exposing Mem0 v2 API for AI agents to store, retrieve, and search long-term memories using semantic search through the standardized MCP protocol.
Overview
Mem0-MCP Server is a self-hosted MCP (Model Context Protocol) server that bridges AI agents with persistent memory storage. It enables intelligent context retention across conversations and sessions using Mem0's AsyncMemory API.
Key Features:
- MCP Protocol Integration - Exposes Mem0 functionality via MCP tools
- Semantic Memory Search - Similarity-based memory retrieval with vector search
- Multi-Tenant Isolation - User/Agent/Session scoped memory isolation
- Flexible Transport - stdio for local agents, SSE for remote connections
- Configuration Management - Pydantic-based validation with environment variable support
Documentation
| Section | Description |
|---|---|
| API Reference | Complete API documentation for all modules and tools |
| Pattern Guides | Design pattern documentation (Singleton, Repository, etc.) |
| Usage Examples | Getting started and advanced usage guides |
| Deployment | Docker Compose configuration and service details |
| Architecture | System architecture and component interactions |
Quick Start
Installation
# Clone and install
git clone https://github.com/your-org/mem0-mcp-server.git
cd mem0-mcp-server
uv sync
# Set environment variables
export OPENAI_API_KEY="your-api-key"
Configuration
Create ~/.config/mem0-mcp-server/settings.json:
{
"vector_store": {
"provider": "redis",
"config": {
"redis_url": "redis://localhost:6379"
}
},
"llm": {
"provider": "openai",
"config": {
"model": "gpt-4o"
}
},
"embedder": {
"provider": "openai",
"config": {
"model": "text-embedding-3-small"
}
}
}
Running the Server
# SSE Transport (remote connections)
uv run python -m mcp_server.main
# stdio Transport (local AI agents)
export MCP_TRANSPORT=stdio
uv run python -m mcp_server.main
OpenCode Configuration
In ~/config/opencode/opencode.json
OpenCode Configuration
In ~/config/opencode/opencode.json
"mcp": {
"mem0": {
"type": "remote",
"enabled": true,
"url": "http://localhost:8050/sse"
}
}
MCP Tools
| Tool | Description |
|---|---|
add_memory | Store information in long-term memory with semantic indexing |
search_memories | Search memories using semantic similarity |
get_memory | Retrieve specific memory by ID |
update_memory | Update existing memory content |
delete_memory | Remove memory from storage |
list_memories | List memories with filtering and pagination |
Usage Example
# Add memory
result = await client.call_tool("add_memory", {
"messages": [{"role": "user", "content": "I prefer dark mode"}],
"user_id": "alice"
})
# Search memories
result = await client.call_tool("search_memories", {
"query": "theme preferences",
"filters": {"user_id": "alice"},
"limit": 5
})
Architecture
AI Agent → FastMCP Server → MemoryManager → Mem0 AsyncMemory → Redis
│ │
├── SafeLogger (stdout/stderr) │
├── Transport (stdio/SSE) │
└── Config (Pydantic validation) │
Components:
- COMP-1: ConfigLoader - Configuration loading and validation
- COMP-2: FastMCP Server - MCP protocol server
- COMP-3: MemoryManager - Memory operations with multi-tenant isolation
- COMP-4: MCP Tools - Tool definitions
- COMP-5: SafeLogger - Output stream separation
Configuration
Parameter Precedence
Configuration values are resolved in order:
- Tool parameters (direct)
- Environment variables (with MCP_ prefix)
- Config file values
- Hardcoded defaults
Environment Variables
| Variable | Default | Description |
|---|---|---|
OPENAI_API_KEY | (required) | OpenAI API key for LLM |
MCP_TRANSPORT | sse | Transport type (stdio, sse) |
MCP_HOST | 0.0.0.0 | Server bind address |
MCP_PORT | 8080 | Server bind port |
Deployment
Docker
# Using docker-compose
docker-compose up -d
# Using Makefile
make docker-up # Start services with docker compose
make docker-down # Stop services
make docker-logs # Show logs
Services:
| Service | Description |
|---|---|
mem0-mcp | MCP server exposing Mem0 API on port 8050 |
ollama-qwen3-embedding | Ollama with qwen3-embedding:8b for vector embeddings (port 11434) |
ollama-qwen | Ollama with qwen2.5:7b for chat completions (port 11435) |
See Deployment → Docker for detailed configuration.
Kubernetes
# Using Helm chart
helm install mem0-mcp ./charts/mem0-mcp-server
Development
| Command | Description |
|---|---|
make install | Install dependencies with uv |
make lint | Lint code with ruff |
make lint-fix | Auto-fix linting issues |
make typecheck | Type check with pyright |
make test | Run all tests |
make test-unit | Run unit tests only |
make test-coverage | Run tests with coverage report |
make build | Build Docker image |
make run | Run development server |
Run multiple commands: make install && make lint && make typecheck && make test
See Makefile for all available commands including Docker management (docker-up, docker-down, docker-logs, etc.).
Project Structure
mem0-mcp/
├── src/mcp_server/
│ ├── __init__.py # FastMCP singleton
│ ├── lifespan.py # Resource lifecycle
│ ├── transport.py # Transport selection
│ ├── memory/
│ │ ├── manager.py # MemoryManager
│ │ └── lifespan.py # AsyncMemory lifecycle
│ ├── config/
│ │ ├── settings.py # Pydantic models
│ │ └── loader.py # Config file loading
│ ├── tools/
│ │ ├── add_memory.py
│ │ ├── search_memories.py
│ │ └── ...
│ └── utils/
│ └── safe_logger.py # Output separation
├── doc/
│ ├── api/ # API reference
│ ├── patterns/ # Pattern guides
│ ├── examples/ # Usage examples
│ └── architecture/ # Architecture docs
├── tests/
├── Makefile
├── Dockerfile
└── docker-compose.yml
License
MIT License
相關伺服器
MemPalace Cloud
Hosted MCP memory across Claude Code, Cursor, ChatGPT and any MCP client. Community-hosted instance of the MIT-licensed MemPalace engine. EU-hosted, GDPR-compliant. Free tier with 200 memories.
Mnemory
A self-hosted, secure, feature-rich memory system for AI agents and assistants. Provides intelligent fact extraction and deduplication, with an artifact store for detailed content.
AgentMemory MCP
Governed persistent memory for AI agents.
VEKTOR Slipstream
[VEKTOR Memory] (https://vektormemory.com) - Local-first persistent memory for AI agents. SQLite-vec, 4-layer associative graph, 8ms recall, 34 tools. No cloud.
Contextful
Highly efficient context management for agentic AI: MCP code search, evidence packs, graph context, and memory for large projects.
Memento
A local-first, LLM-agnostic memory layer for AI assistants
OpenMemBrain
OpenMemBrain is the intelligent membrane for AI coding memory. It autonomously reads and learns from your coding sessions — you never have to tell it what to save. It selectively absorbs project knowledge, blocks secrets, filters noise, resolves conflicts, and persists only what matters.
Adaptive Recall
Adaptive MCP memory system for AI applications. Learns which retrieval strategies work for your data, scores results using cognitive science models, builds a knowledge graph automatically, and validates every parameter change against real query history before adopting it. Patent pending.
Agentmemory MCP
Paid remote MCP endpoint for persistent AI agent memory, site analytics, pricing, checkout status, and search-readiness operations.
Link
Local Markdown memory for AI agents