Nocturne Memory

Active
GitHub Python MIT

Description

A lightweight, rollbackable, and visual long-term memory server for MCP agents, replacing traditional vector RAG with reliable context retention.

Key Features

  • MCP-based long-term memory server enabling AI to remember identity and history across sessions and models
  • One soul, any engine — memory stored in independent MCP Server, not bound to any LLM, AI retains full memory when switching models
  • Namespace isolation support — multiple AI personalities can have completely independent memory spaces without interference
  • Visual Dashboard management panel with tree-view browsing, real-time editing, auditing and version rollback
  • Automatic backup on every AI operation, cleanup requires human confirmation, providing version safety net
  • Compatible with all MCP-supporting clients: Claude Code, Cursor, Windsurf, GitHub Copilot, Gemini CLI and more

Use Cases

💡 Enable AI assistants to remember user preferences, work context, and personal history across sessions without re-explaining
💡 Maintain AI personality consistency and memory continuity when switching between Claude and Gemini
💡 Build AI companions with independent personalities, accumulating emotional memory and interaction history over time
💡 Multiple AI agents share the same memory infrastructure while maintaining independent memory spaces
💡 Audit and manage all AI memories through visual panel ensuring privacy and security

Tags

python mcp memory agent tools

Categories

🧩 Agent Memory

Quick Start

Install: git clone https://github.com/Dataojitori/nocturne_memory.git && cd nocturne_memory && pip install -r backend/requirements.txt. Add to AI client MCP config: {"mcpServers": {"nocturne_memory": {"command": "python", "args": ["/path/to/backend/mcp_server.py"]}}. Dashboard auto-builds on first start. Verify: tell AI "Read system://boot. Tell me who you are."
Visit GitHub

Related Projects