# Mnemosyne

> Mnemosyne is an open-source MCP server for knowledge and memory. Local-first agent memory in one SQLite file; corrections retire dependent memories. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted, API
- **Website:** https://mnemosy.ai
- **Repository:** https://github.com/28naem-del/mnemosyne
- **MCP Registry name:** io.github.28naem-del/mnemosyne
- **Categories:** knowledge and memory, AI models and LLM tooling
- **Listing updated:** 2026-11-14

## About

Mnemosyne tackles stale agent memory. Every record keeps its source and trust tier, and when a source is corrected, the procedures and skills recorded as depending on it are retired, so the next agent starts from evidence still supported. The 3.0 pre-release can also propose dependencies nobody declared, for review. The superseded version stays inspectable, and past beliefs can be replayed exactly.

The local engine needs no model API key, database server or Docker, and it can run over MCP, a TypeScript SDK, a CLI or authenticated HTTP. The site pairs its numbers with their limits, such as 81.8% on a 500-question memory benchmark read by Claude Sonnet 4.6 and judged by Claude Haiku 4.5, where the prompt was tuned on 100 of those questions.

This is the TypeScript package mnemosy-ai, MIT-licensed on Node.js 22.16 or newer, with 2.0.0 stable and 3.0.0 release candidates on the next tag. The README stresses it is a separate project from the Python mnemosyne-memory package.

## Features

- **Evidence on every memory:** Each record keeps its source, trust tier and superseded versions.
- **Correction cascades:** Fixing a source retires memories recorded as depending on it.
- **Replayable history:** Past beliefs replay exactly and a bisect tool finds records that displaced others.
- **Skill trials:** Candidate skills need passed trials before entering ordinary advice.
- **Several hosts:** MCP, TypeScript SDK, CLI and authenticated HTTP.
- **Local backups:** Verified local backups let you rehearse recovery.

## Good for

- Give a coding agent memory that survives sessions
- Stop an agent reusing a lesson based on a corrected requirement
- Audit why a retrieval missed a record
- Carry decisions and next steps into the next session
- Run agent memory offline without an API key

## Install

### Claude Code (Terminal)

```sh
claude mcp add mnemosyne -- npx -y mnemosy-ai
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "mnemosyne": {
      "command": "npx",
      "args": [
        "-y",
        "mnemosy-ai"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "mnemosyne": {
      "command": "npx",
      "args": [
        "-y",
        "mnemosy-ai"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "mnemosyne": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "mnemosy-ai"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "mnemosyne": {
      "command": "npx",
      "args": [
        "-y",
        "mnemosy-ai"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "mnemosyne": {
      "command": "npx",
      "args": [
        "-y",
        "mnemosy-ai"
      ]
    }
  }
}
```

## FAQ

### Does it need an API key or database?

No. The README says the local engine needs no model API key, database server or Docker.

### Which version should I install?

2.0.0 is stable on npm latest. The 3.0.0 release candidate is on npm next.

### Is it the same as the Python mnemosyne project?

No. The README says they share a name only.

### What licence and runtime?

MIT, with Node.js 22.16 or newer.

### Is Mnemosyne free?

Yes. Mnemosyne is open source, so you can use and self-host it for free.

### How do I add Mnemosyne to Claude Code?

Run this command in your terminal: claude mcp add mnemosyne -- npx -y mnemosy-ai

### What are the best alternatives to Mnemosyne?

The closest alternatives to Mnemosyne are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

## Alternatives

- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Nautilus Compass Memory](https://ai-powered.com/mcp/nautilus-compass-memory): Nautilus Compass Memory is an open-source MCP server for knowledge and memory.
- [note1](https://ai-powered.com/mcp/note1): note1 is an MCP server for productivity.
- [MedXpert 图书馆连接器（通用版）](https://ai-powered.com/mcp/medxpert): MedXpert 图书馆连接器（通用版） is an open-source MCP server for knowledge and memory.
- [Context Nest](https://ai-powered.com/mcp/context-nest): Context Nest is a freemium MCP server for knowledge and memory.
- [Tulsa Ecosystem Map](https://ai-powered.com/mcp/tulsa-ecosystem-map): Tulsa Ecosystem Map is a free MCP server for knowledge and memory.

---
Source: https://ai-powered.com/mcp/mnemosyne-2
