nhadaututtheky-neural-memory

TotalClaw 作者 totalclaw

类脑反射式记忆系统:扩散激活召回、认知推理与多设备同步,含 45 个 MCP 工具。

安装 / 下载方式

TotalClaw CLI推荐
totalclaw install totalclaw:totalclaw~nhadaututtheky-neural-memory
cURL直接下载,无需登录
curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~nhadaututtheky-neural-memory/file -o nhadaututtheky-neural-memory.md
## 概述(中文)

类脑反射式记忆系统:扩散激活召回、认知推理与多设备同步,含 45 个 MCP 工具。

## 技能正文

# Neural Memory(神经记忆)

面向 AI 智能体的反射式记忆系统——将经验存储为互联神经元,通过扩散激活召回,模拟人脑工作方式。

## 功能

Neural Memory 为 AI 智能体提供跨会话的持久、联想记忆。不用关键词搜索,而是通过神经图的扩散激活——一起激活的记忆会联结在一起。

## 关键特性

- **45 MCP tools** for persistent memory + cognitive reasoning
- **Spreading activation recall** — not keyword search, memories activate related memories
- **Cognitive reasoning** — hypotheses, evidence, predictions, schema evolution
- **Knowledge base training** from PDF, DOCX, PPTX, HTML, JSON, XLSX, CSV
- **Multi-device sync** with neural-aware conflict resolution
- **4 embedding providers** — Sentence Transformers, Gemini, Ollama, OpenAI
- **Retrieval pipeline** — RRF score fusion, graph expansion, Personalized PageRank
- **Session intelligence** — topic EMA tracking, LRU eviction, auto-expiry
- **React 仪表板** — 7 pages: health, evolution, graph, timeline, settings
- **VS Code extension** — status bar, graph explorer, CodeLens, memory tree
- **Fernet encryption** for sensitive content
- **Brain versioning** — snapshots, rollback, export/import
- **Telegram backup** — send brain .db to chat/group/channel

## 安装

```bash
pip install neural-memory
```

含嵌入模型:

```bash
pip install neural-memory[embeddings]
```

## MCP 配置

```json
{
  "mcpServers": {
    "neural-memory": {
      "command": "uvx",
      "args": ["--from", "neural-memory", "nmem-mcp"]
    }
  }
}
```

## 用法

配置后 Neural Memory 自动工作。

**召回(RECALL)** — 回应涉及过往工作的任务前:
- New session → `nmem_recall("current project context")`
- Past decision/event → `nmem_recall("<project> <topic>")`
- Skip for purely new, self-contained questions

**保存(SAVE)** — 完成每项任务后,若做出决策、修复 bug、学到偏好或发现模式:
- `nmem_remember(content="Chose X over Y because Z", type="decision", priority=7, tags=["project", "topic"])`
- Use causal language (not flat facts). Max 1-3 sentences.
- Do NOT save ephemeral file reads, things in git history, or duplicates.

**刷新(FLUSH)** — 会话结束时:
- `nmem_auto(action="process", text="brief summary")`

## 记忆类型

| 类型 | 用途 |
|------|---------|
| fact | 稳定知识 |
| decision | 「因 Z 选择 X 而非 Y」 |
| insight | 发现的模式 |
| error | Bug 与根因 |
| workflow | 流程步骤 |
| preference | 用户偏好 |
| instruction | 需遵循的规则 |

## 链接

- [GitHub](https://github.com/nhadaututtheky/neural-memory)
- [Documentation](https://nhadaututtheky.github.io/neural-memory)
- [VS Code Extension](https://marketplace.visualstudio.com/items?itemName=neuralmem.neuralmemory)