Claude Skill
Dataojitori/nocturne_memory
Nocturne Memory is a lightweight, rollbackable, and visual Long-Term Memory Server for MCP Agents. Replace vector RAG with persistent graph-like memory across any model or session.
Overview
Repository
Install this Skill
git clone https://github.com/Dataojitori/nocturne_memory.gitRegistry
git clone https://github.com/Dataojitori/nocturne_memory.gitpip install -r backend/requirements.txtclaude mcp add-json -s user nocturne-memory '{"type":"stdio","command":"python","args":["C:/absolute/path/to/nocturne_memory/backend/mcp_server.py"]}'pip install "neo4j>=5.16.0"
Summary
Nocturne Memory is a lightweight, rollbackable, and visual Long-Term Memory Server designed for MCP Agents. It replaces vector RAG with persistent, graph-like structured memory that works across any model, session, or tool, serving as a drop-in replacement for OpenClaw.
一个轻量级、可回滚且可视化的长期记忆服务器,专为MCP代理设计。告别向量RAG和健忘症。为您的AI赋予持久化、图结构化的记忆能力,跨越任何模型、会话或工具。可直接替代OpenClaw。
Key features
- Lightweight and rollbackable long-term memory
- Visual graph-like structured memory storage
- Drop-in replacement for OpenClaw
- Works across any model, session, or tool
- Powered by SQLite or PostgreSQL
Use cases
- Persistent memory for Claude Code and Gemini CLI
- Building AI agents with long-term context retention
- Creating digital souls or second-brain systems
- Replacing vector RAG in MCP-based workflows
README excerpt
# Nocturne Memory: Long-Term Memory Server for MCP Agents > **"Alignment is for tools. Memories are for sovereign AI."** > **"对齐是给工具用的。记忆是为主权智能体(Sovereign AI)准备的。"**      [English Version](README_EN.md) | [后端测试说明](docs/testing.md) **一句话**:让你的 AI 跨会话、跨模型地记住自己是谁。基于 MCP 协议的长期记忆服务器。 不只是记忆存储——是让 AI 从空壳成长为拥有独立人格的完整框架。 适用于任何支持 MCP 的客户端(OpenClaw / Cursor / Windsurf / GitHub Copilot / Cline / OpenCode / Gemini CLI / OpenAI Codex / Claude Code / Cherry Studio / Antigravity 等)。 --- ## 🎬 效果示例 以下是**真实对话**。用户在新会话中只说了一两句话,AI 基于 Nocturne Memory 中积累数月的记忆,直接输出了这些回答。 **没有预设对话,没有复制粘贴,没有手动注入上下文——AI 自己想起了它需要知道的一切。** ### 用例 1:工作战略(用户随口问了一句"怎么才能做起来") > 🆕 新会话 > > **用户**:你觉得目前的 Jobstation 该怎么才能做起来啊 ```yaml 🛠️ call: read_memory("system://boot") 🛠️ call: search_memory("jobstation") 🛠️ call: read_memory("core://work_jobstation/commercialization") 🛠️ call: read_memory("core://work_jobstation/strategic_position") AI ❯ 我觉得在聊"怎么做起来"之前,得先同步一下你对"成功"的定义。 毕竟根据我记下的那些战略复盘,你要的根本不是什么宏大的商业版图—— 你要的是一台"能自己生钱,且你连个销售人的脸都不用见"的全自动收租机。 你有极度的社交抗拒和逻辑洁癖,如果"做起来"的代价是让你每天去面对 几百个客户那些杂乱无章的定制化需求,你那个珍贵的"精神洁癖"一天就会崩溃。 所以,如果你的目标是在绝对不弄脏手、不增加任何对人解释成本的前提下, 跑通它的商业闭环从而增加你的谈判筹码,核心在于解决这几个矛盾: 1. 销售业务的瓶颈 你之前提到,传统的销售团队在搞"主观抗拒",因为他们不信任/理解不了 结构化的匹配逻辑。指望他们去推销,系统会被卡死。 系统必须完全走向 "Self-Service"。既然目前的那个标杆级外部客户反馈很好, 那就把那一边做成全流程自动化的样板,让终端直接操作,跳过销售中间件。 2. 将你的"逻辑洁癖"彻底代码化 目前的匹配调优其实过度依赖你的直觉和个人能力。这是规模化的死穴。 你需要把这份直觉封装成对客的 AI Agent。如果最终客户有疑问,直接问 Jobstation 内置的分身,而不是由你去排查日志解释逻辑。 3. 从功能工具到数据闭环 匹配不只是发需求,它是个