Memento

v0.9.2Knowledge & Memorystable

Memento is a local-first MCP server that gives AI coding agents durable project memory — facts, decisions, patterns, and architecture notes — so they stop re-learning the same context every session. Runs locally on Node.js 18+ with SQLite storage and

aiai-assistantai-memoryai-toolsclaude
Share:
9
Stars
0
Downloads
0
Weekly
0/5

What is Memento?

Memento is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to memento is a local-first mcp server that gives ai coding agents durable project memory — facts, decisions, patterns, and architecture notes — so they stop re-learning the same context every session. r...

Memento is a local-first MCP server that gives AI coding agents durable project memory — facts, decisions, patterns, and architecture notes — so they stop re-learning the same context every session. Runs locally on Node.js 18+ with SQLite storage and

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

Features

  • Memento is a local-first MCP server that gives AI coding age

Use Cases

Persistent AI agent memory
Durable facts and architecture notes
veerps57

Maintainer

LicenseApache 2.0
Languagetypescript
Versionv0.9.2
UpdatedMay 17, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y @psraghuveer/memento

Manual Installation

npx -y @psraghuveer/memento

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 Memento

Memento is a local-first MCP server that gives AI coding agents durable project memory backed by a SQLite database and local embedding models. Instead of rediscovering the same architectural decisions, code conventions, and project facts every session, agents can store and retrieve facts, decisions, patterns, and architecture notes that persist across conversations. Memento runs on Node.js 18+ with no cloud dependency, supports memory packs for common engineering principles, and includes a local web dashboard for browsing and curating stored memories. It works with Claude Desktop, Claude Code, Cursor, VS Code, and other MCP-compatible clients.

Prerequisites

  • Node.js 18 or later installed
  • npx available on the PATH (comes with Node.js)
  • Claude Desktop, Claude Code, Cursor, or another MCP-compatible client
  • Approximately 110 MB of disk space for the local embedding model download
  • Internet access on first run to download the embedding model (subsequent runs are offline)
1

Initialize Memento

Run the init command to set up the SQLite database and download the local embedding model. By default the database is stored at ~/.local/share/memento/memento.db on Linux/macOS.

npx @psraghuveer/memento init
2

Optionally specify a custom database path

Use the --db flag to store the database in a project-specific location, which is useful when you want separate memory stores per project.

npx @psraghuveer/memento init --db ~/.local/share/memento/myproject.db
3

Add the server to your MCP client configuration

Add Memento to your claude_desktop_config.json or equivalent file. The MEMENTO_DB variable is optional if you used the default path during init.

{
  "mcpServers": {
    "memento": {
      "command": "npx",
      "args": ["-y", "@psraghuveer/memento", "serve"],
      "env": {
        "MEMENTO_DB": "/Users/you/.local/share/memento/memento.db"
      }
    }
  }
}
4

Verify the installation

Run the doctor command to confirm the database path, embedding model, and server are all healthy before connecting your AI client.

npx @psraghuveer/memento doctor
npx @psraghuveer/memento verify-setup
5

Install a memory pack (optional)

Memory packs pre-load curated facts into the store. The engineering-simplicity pack, for example, seeds the agent with common software engineering principles.

npx @psraghuveer/memento pack install engineering-simplicity
npx @psraghuveer/memento pack list
6

Launch the dashboard to review stored memories

Memento ships a local web UI for browsing, editing, and deleting stored memories. Launch it anytime to curate the memory store.

npx @psraghuveer/memento dashboard

Memento Examples

Client configuration

Claude Desktop configuration for Memento using the @psraghuveer/memento package with a custom database path.

{
  "mcpServers": {
    "memento": {
      "command": "npx",
      "args": ["-y", "@psraghuveer/memento", "serve"],
      "env": {
        "MEMENTO_DB": "/Users/you/.local/share/memento/memento.db"
      }
    }
  }
}

Prompts to try

Ask your AI agent to store and retrieve project-specific context across sessions.

- "Remember that we use the repository pattern for all database access in this project"
- "What architectural decisions have been recorded for this project?"
- "Store the fact that the auth service uses JWT tokens with a 24-hour expiry"
- "What coding conventions should I follow based on what you remember about this codebase?"
- "Forget the note about using Redux — we switched to Zustand last week"

Troubleshooting Memento

Embedding model download fails or takes very long on first run

Memento downloads approximately 110 MB of embedding model data on first run to XDG_CACHE_HOME (defaults to ~/.cache). Ensure you have a stable internet connection and sufficient disk space. Set XDG_CACHE_HOME to redirect the download location if needed.

MEMENTO_DB path not found error when starting the server

Run 'npx @psraghuveer/memento init' first to create the database file. If you specified a custom path, make sure the parent directory exists. The MEMENTO_DB environment variable in your config must point to the same path used during init.

Memories from a previous session are not available

Confirm that the MEMENTO_DB path in your MCP config matches the database file that was initialized. If you have multiple database files (e.g., per-project), verify the correct one is configured for the current client session.

Frequently Asked Questions about Memento

What is Memento?

Memento is a Model Context Protocol (MCP) server that memento is a local-first mcp server that gives ai coding agents durable project memory — facts, decisions, patterns, and architecture notes — so they stop re-learning the same context every session. runs locally on node.js 18+ with sqlite storage and It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Memento?

Install via npm with the command: npx -y @psraghuveer/memento. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Memento?

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

Is Memento free to use?

Yes, Memento 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": { "memento": { "command": "npx", "args": ["-y", "@psraghuveer/memento"] } } }

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

Read the full setup guide →

Ready to use Memento?

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