Extended Memory

Provides cross-session memory storage for Claude Desktop with project isolation and local storage.

python

14

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ssmirnovpro-extended-memory-mcp": {
      "command": "python3",
      "args": [
        "-m",
        "extended_memory_mcp.server"
      ],
      "env": {
        "LOG_LEVEL": "INFO",
        "STORAGE_CONNECTION_STRING": "sqlite:///C:/path/to/memory.db"
      }
    }
  }
}

Extended Memory MCP is a server that provides persistent, cross-session memory storage for Claude Desktop. It saves and restores your project context, architectural decisions, communication style, and working preferences, while keeping data isolated between different projects so you can resume work seamlessly across conversations.

How to use

You connect Claude Desktop to the Extended Memory MCP server to give Claude access to your project memory. The server is run locally and communicates with Claude through the Model Context Protocol. You can save important project details, load them in later sessions, and manage memory per project so nothing spills between unrelated work.

How to install

Prerequisites you need before installation are clear: you must have Python 3.8 or newer and the Desktop Claude app installed on your machine.

Step 1: Install the MCP package

pip install extended-memory-mcp

Step 2: Run the MCP server locally. The project provides a standard Python module entry point suitable for the Desktop Claude integration.

python3 -m extended_memory_mcp.server

Configuration for the Desktop Claude app

Add the Extended Memory MCP server to your MCP configuration in the Desktop Claude app. This enables Claude to communicate with the local memory store.

{
  "mcpServers": {
    "extended-memory": {
      "command": "python3",
      "args": ["-m", "extended_memory_mcp.server"],
      "env": {
        "LOG_LEVEL": "INFO"
      }
    }
  }
}

Configuration for source installation

If you installed from the source path, use the explicit path to the server module in the configuration.

{
  "mcpServers": {
    "extended-memory": {
      "command": "python3",
      "args": ["/path/to/extended-memory-mcp/mcp-server/server.py"],
      "env": {
        "LOG_LEVEL": "INFO"
      }
    }
  }
}

Additional installation options

Optional features and development options are available to tailor the server to your setup.

pip install extended-memory-mcp[redis]

pip install extended-memory-mcp[dev]

Configuration parameters

Use these environment variables to customize storage and logging.

STORAGE_CONNECTION_STRING: Database location
LOG_LEVEL: Logging verbosity

Platform notes and storage options

macOS & Linux use the default storage location in the user data directory if you don’t set a custom path.
Windows users should specify STORAGE_CONNECTION_STRING to point to a local SQLite file.
Redis storage is available as an experimental option by setting STORAGE_CONNECTION_STRING to a Redis URL.

"env": {
  "STORAGE_CONNECTION_STRING": "sqlite:///C:/Users/YourName/extended-memory/memory.db"
}

Verification

To verify the setup, restart Claude Desktop, confirm that the Extended Memory MCP server appears in Developer Settings, and run a quick memory test by asking Claude to save a memory for a project. Then start a new conversation and check that the memory persists.

Privacy & Local Storage

Your memory data stays local to your machine. There is no cloud sync, no data processing, and no telemetry. You maintain full control over all stored information.

Project Organization & Usage

Memory is organized by projects with strict isolation. If you don’t specify a project_id, conversations default to the general scope. To ensure proper project isolation, include the project identifier in Claude’s custom instructions when reading or writing memory.

Example instruction to include in Claude: You have external memory; always specify that this is project "<your-short-project-id-here>" as its project_id for read and write operations.

Replace the placeholder with your actual project identifier, such as mobile_app, blog_redesign, or work_project.

Memory tools you can use

Extended Memory provides Claude with several memory management tools. Claude may choose when to load memory, but you can explicitly trigger actions when needed.

Examples you can use with Claude include commands to load, save, or manage memory.

Security & Privacy

All data remains local to your device. There is no telemetry or external data collection.

License

MIT License. Free to use, modify, and distribute.

Available tools

save_context

Save important information with tags to memory

load_contexts

Load previous context and conversations into the current session

forget_context

Remove outdated or no longer relevant memory from the store

list_all_projects

View all projects with their memory contexts and metadata

get_popular_tags

Find commonly used tags to organize memory