nocturne_memory

概述

Nocturne Memory 是一个先进的 MCP 服务器，为 AI 代理提供跨模型、跨会话和跨工具的持久化结构化记忆。它用类图记忆结构替代传统向量 RAG，AI 可以创建、更新和组织自己的记忆，并具有版本控制和条件触发功能。服务器配备可视化仪表板，供人类审查和管理记忆，确保 AI 自主性的同时保持监督。与传统仅绑定单一平台的记忆系统不同，Nocturne Memory 实现'一个灵魂，任意引擎'——允许代理在不同 LLM 间切换时保持身份一致性。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• read_memory

  Read a specific memory node by URI

  试试:Read my core identity试试:Show me memories about work project试试:Read system://boot
• search_memory

  Search for memories containing specific keywords

  试试:Search for memories about 'Jobstation'试试:Find memories containing 'strategy'试试:Search for memories about user preferences
• create_memory

  Create a new memory node with content and metadata

  试试:Create a new memory about today's conversation试试:Add a memory about my preferences for coffee试试:Create a memory note about project deadlines
• update_memory

  Update an existing memory node with new content

  试试:Update my core identity with new information试试:Modify the memory about yesterday's meeting试试:Update the memory about my favorite books
• delete_memory

  Delete a memory node

  试试:Delete the memory about outdated information试试:Remove the temporary memory from yesterday试试:Delete the memory that's no longer relevant
• list_memories

  List all memories under a specific domain

  试试:List all memories in my core domain试试:Show all memories in the project namespace试试:List memories in the system domain
• get_memory_index

  Get an index of all memories in a specific domain

  试试:Get index of all core memories试试:Show me the index of all work-related memories试试:Get index of memories in the project namespace

可对比工具

安装

git clone https://github.com/Dataojitori/nocturne_memory.git
cd nocturne_memory
pip install -r backend/requirements.txt

{
  "mcpServers": {
    "nocturne_memory": {
      "command": "python",
      "args": ["/path/to/nocturne_memory/backend/mcp_server.py"]
    }
  }
}

{
  "mcpServers": {
    "nocturne-memory": {
      "command": "python",
      "args": ["/path/to/nocturne_memory/backend/mcp_server.py"]
    }
  }
}

{
  "mcpServers": {
    "nocturne_memory": {
      "command": "python",
      "args": ["/path/to/nocturne_memory/backend/mcp_wrapper.py"]
    }
  }
}

FAQ

Nocturne Memory 与传统向量 RAG 系统有何不同？

向量 RAG 系统将知识切碎成向量，丢失结构并使 AI 成为被动接收者。Nocturne Memory 通过基于 URI 的路由保留层次结构，允许 AI 主动创建和更新记忆并具有版本控制，并实现精确的条件触发而非盲目的相似性搜索。

我可以同时使用多个 AI 人格吗？

是的，Nocturne Memory 支持命名空间隔离，允许您运行多个独立的记忆空间，为不同的 AI 人格（如 Alice 和 Bob）使用，互不干扰。

nocturne_memory 对比