MCP Memory Service

v1.0.0Knowledge & Memorystable

Provides semantic memory and persistent storage for Claude, leveraging ChromaDB and sentence transformers for enhanced search and retrieval capabilities.

agent-memoryagentic-aiai-agentsautogenclaude
Share:
1,867
Stars
0
Downloads
0
Weekly
0/5

What is MCP Memory Service?

MCP Memory Service is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to provides semantic memory and persistent storage for claude, leveraging chromadb and sentence transformers for enhanced search and retrieval capabilities.

Provides semantic memory and persistent storage for Claude, leveraging ChromaDB and sentence transformers for enhanced search and retrieval capabilities.

This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Provides semantic memory and persistent storage for Claude,

Use Cases

Store and retrieve semantic memories for Claude using ChromaDB.
Enable enhanced search and retrieval with sentence transformers.
Persist knowledge across conversation sessions.
doobidoo

Maintainer

LicenseApache 2.0
Languagepython
Versionv1.0.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx mcp-memory-service

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use MCP Memory Service

MCP Memory Service is an open-source persistent memory backend for AI agents that stores and retrieves semantic memories across separate conversation sessions using local ONNX sentence-transformer embeddings and a ChromaDB-compatible vector store. It supports hybrid search combining BM25 full-text and vector similarity, a knowledge graph with typed relation edges, autonomous memory consolidation with decay scoring, and a web dashboard for browsing stored knowledge. Developers use it to give Claude and other agents durable long-term memory without relying on any cloud memory provider.

Prerequisites

  • Python 3.10+ installed on your system
  • pip for installing the mcp-memory-service package
  • An MCP-compatible client such as Claude Desktop
  • Optional: an OpenAI-compatible local LLM endpoint (for quality scoring, not required for basic use)
  • Sufficient disk space for the local embedding model (all-MiniLM-L6-v2 is downloaded on first run)
1

Install the package via pip

Install mcp-memory-service and all its dependencies including the sentence-transformers ONNX runtime.

pip install mcp-memory-service
2

Start the memory server

Run the server using the installed 'memory' command. On first start it downloads the default embedding model (all-MiniLM-L6-v2) and initializes the local vector store.

memory server
3

Configure environment variables (optional)

Set environment variables to customize the embedding model, enable anonymous access for local development, or switch to a multilingual embedding model.

export MCP_ALLOW_ANONYMOUS_ACCESS=true
export MCP_EMBEDDING_MODEL=all-MiniLM-L6-v2
# For multilingual support:
export MCP_EMBEDDING_MODEL=paraphrase-multilingual-MiniLM-L12-v2
4

Add to Claude Desktop configuration

Register the memory MCP server in Claude Desktop's config file so Claude can call memory tools during conversations.

5

Restart Claude Desktop

Quit and relaunch Claude Desktop for the new MCP server configuration to take effect. Memory tools will now be available in every conversation.

MCP Memory Service Examples

Client configuration

Add this to your Claude Desktop config file at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).

{
  "mcpServers": {
    "memory": {
      "command": "memory",
      "args": ["server"],
      "env": {
        "MCP_ALLOW_ANONYMOUS_ACCESS": "true",
        "MCP_EMBEDDING_MODEL": "all-MiniLM-L6-v2"
      }
    }
  }
}

Prompts to try

Once the memory server is running and connected, use these prompts to store and retrieve information across sessions.

- "Remember that our production database uses PostgreSQL 15 with read replicas in us-east-1"
- "What do you remember about our database setup?"
- "Store the fact that the payment service API key rotates every 90 days"
- "Search your memory for anything related to API authentication in this project"
- "Create a relation: the user-service depends on the auth-service"
- "Retrieve all memories tagged with 'architecture' from this project"

Troubleshooting MCP Memory Service

Server fails to start with model download errors

The sentence-transformer model is downloaded from Hugging Face on first run. Ensure you have an internet connection and at least 500 MB of free disk space. If behind a proxy, set HTTPS_PROXY environment variable. Retry by running 'memory server' again after connectivity is restored.

Claude does not show memory tools after adding the config

Verify that 'memory' is on your PATH by running 'which memory' or 'memory --version'. If the command is not found, the pip install target bin directory is not on PATH. Add it: run 'python -m site --user-base' and add its 'bin' subdirectory to PATH, then restart Claude Desktop.

Semantic search returns irrelevant results

The default all-MiniLM-L6-v2 model is tuned for English short sentences. For long documents or multilingual content, switch to paraphrase-multilingual-MiniLM-L12-v2 by setting MCP_EMBEDDING_MODEL. If memories were stored with a different model, clear and re-index the memory store after switching models.

Frequently Asked Questions about MCP Memory Service

What is MCP Memory Service?

MCP Memory Service is a Model Context Protocol (MCP) server that provides semantic memory and persistent storage for claude, leveraging chromadb and sentence transformers for enhanced search and retrieval capabilities. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install MCP Memory Service?

Follow the installation instructions on the MCP Memory Service GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with MCP Memory Service?

MCP Memory Service works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is MCP Memory Service free to use?

Yes, MCP Memory Service is open source and available under the Apache 2.0 license. You can use it freely in both personal and commercial projects.

Browse More Knowledge & Memory MCP Servers

Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "mcp-memory-service": { "command": "npx", "args": ["-y", "mcp-memory-service"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use MCP Memory Service?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides