tech-memory-mcp
A MCP server that gives Claude Code and other AI assistants long-term memory by automatically extracting technical knowledge from conversations and retrieving relevant experiences in future sessions.
README
tech-memory-mcp
不只是记住"说了什么",而是记住"学到了什么"。
让 Claude Code 等 AI 编程助手具备长期记忆能力,从每次技术对话中自动提取知识点,下次对话时自动检索相关经验。
核心特性
- 🧠 自动提取:从对话中识别 decision/lesson/preference/fact/pattern 五类知识
- 🔍 混合检索:向量语义搜索 + FTS5 全文搜索,中英文混合查询
- 🇨🇳 中文优先:基于 jina-embeddings-v2-base-zh(768 维)嵌入模型
- 📦 完全本地:SQLite + Transformers.js,无需外部服务
- 🔄 跨项目累积:所有项目知识存入同一数据库,形成终身技术积累
系统要求
- Node.js >= 18.0.0
- 首次启动下载嵌入模型约 154MB
- Windows 用户:如安装失败,请先运行
npm install --global windows-build-tools
快速开始
第一步:安装
方式一:npx 安装(推荐)
npx -y tech-memory-mcp@latest
首次运行会自动下载嵌入模型(约 154MB,30-60 秒)。模型仅需下载一次,后续启动直接使用缓存。
方式二:从源码安装
git clone https://github.com/myx0423/tech-memory-mcp.git
cd tech-memory-mcp
npm install
npm run build
然后将 .mcp.json 配置改为:
{
"mcpServers": {
"tech-memory": {
"command": "node",
"args": ["/你的路径/tech-memory-mcp/dist/index.js"]
}
}
}
第二步:配置 Claude Code
在项目根目录或 ~/.claude/ 创建 .mcp.json:
{
"mcpServers": {
"tech-memory": {
"command": "npx",
"args": ["-y", "tech-memory-mcp"]
}
}
}
第三步:验证安装
启动 Claude Code,输入:
请调用 tech_stats 查看知识库统计
如果返回知识库统计信息(即使为空),说明安装成功。首次启动会自动下载模型,约 30-60 秒。
国内网络加速
如果网络环境不佳,可以提前手动下载模型:
1. 下载模型文件
从 HuggingFace 镜像站下载模型到本地目录:
mkdir -p ~/models/jina-embeddings-v2-base-zh
cd ~/models/jina-embeddings-v2-base-zh
# 使用 wget 下载(或使用浏览器手动下载)
wget https://hf-mirror.com/Xenova/jina-embeddings-v2-base-zh/resolve/main/config.json
wget https://hf-mirror.com/Xenova/jina-embeddings-v2-base-zh/resolve/main/model.onnx
wget https://hf-mirror.com/Xenova/jina-embeddings-v2-base-zh/resolve/main/tokenizer.json
wget https://hf-mirror.com/Xenova/jina-embeddings-v2-base-zh/resolve/main/vocab.txt
2. 配置环境变量
修改 .mcp.json,添加 TECH_MEMORY_MODEL_PATH:
{
"mcpServers": {
"tech-memory": {
"command": "npx",
"args": ["-y", "tech-memory-mcp"],
"env": {
"TECH_MEMORY_MODEL_PATH": "/Users/yourname/models/jina-embeddings-v2-base-zh"
}
}
}
}
Windows 用户示例:
{
"mcpServers": {
"tech-memory": {
"command": "npx",
"args": ["-y", "tech-memory-mcp"],
"env": {
"TECH_MEMORY_MODEL_PATH": "C:\\Users\\yourname\\models\\jina-embeddings-v2-base-zh"
}
}
}
}
可选:自定义模型缓存目录(用于自动下载的模型):
{
"mcpServers": {
"tech-memory": {
"command": "npx",
"args": ["-y", "tech-memory-mcp"],
"env": {
"TECH_MEMORY_CACHE_DIR": "/path/to/cache"
}
}
}
}
MCP 工具
| 工具 | 作用 |
|---|---|
tech_search |
混合搜索(语义 + 全文),支持置信度过滤和过期知识过滤,自动记录曝光事件 |
tech_store |
存储知识点,自动去重,支持置信度和过期时间 |
tech_extract_template |
返回中文知识提取 Prompt + JSON Schema |
tech_auto_extract |
自动从对话中提取并存储技术知识(支持两阶段调用) |
tech_confirm |
确认知识点是否有用,调整置信度,记录采用/拒绝事件 |
tech_decay |
执行知识老化衰减,批量更新置信度,清理过期使用事件 |
tech_outdated |
查询过期或低置信度的知识点 |
tech_conflict_scan |
扫描知识库中的冲突和重复条目 |
tech_resolve |
处理冲突条目(保留、合并或标记矛盾) |
tech_link |
创建知识点之间有向关系 |
tech_get |
查询知识点 + 关联条目 |
tech_stats |
数据库统计 |
tech_usage_stats |
知识库使用统计(曝光、采用、采用率、热门查询) |
tech_export |
导出知识库为 Markdown 或 JSON 格式 |
tech_import |
从 JSON/Markdown/纯文本批量导入知识 |
tech_backup |
创建 SQLite 数据库完整备份 |
知识分类
| 类型 | 说明 | 示例 |
|---|---|---|
decision |
技术决策 | "选了 Calico 而非 Flannel,因为需要 NetworkPolicy" |
lesson |
经验教训 | "kubeadm init 失败记得先 kubeadm reset -f" |
preference |
个人偏好 | "更喜欢用 Containerd 而非 Docker 作为运行时" |
fact |
技术事实 | "K8s v1.34 对应 Containerd 沙箱镜像 pause:3.10" |
pattern |
通用模式 | "三 Master + Nginx + KeepAlived 是生产标配" |
工作原理
对话结束 → tech_auto_extract 提取知识点
→ tech_store 存储(自动嵌入 + 语义去重)
→ 下次对话时 tech_search 自动检索相关记忆
→ tech_confirm 反馈知识价值,优化搜索排序
数据存储
- 数据库:
~/.tech-memory/memory.db(SQLite + 内置向量搜索) - 嵌入模型:
Xenova/jina-embeddings-v2-base-zh(q8 量化,154 MB) - 完全本地,无需任何外部服务
开发
git clone https://github.com/myx0423/tech-memory-mcp.git
cd tech-memory-mcp
npm install
npm run build
node dist/index.js # 启动 MCP Server
运行测试:
npm test
姊妹项目
| 项目 | 说明 |
|---|---|
| remote-ops-mcp | SSH 远程服务器管理:执行命令、部署服务、排查故障。操作失败时自动缓存踩坑上下文,配合 tech_memory.tech_store 形成运维知识闭环。 |
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.