notes-mcp
MCP server for a personal notes/knowledge base that enables AI assistants to create, search, and retrieve notes using natural language. It exposes tools like create_note and search_notes, resources for each note, and a summarize_tag prompt.
README
notes-mcp
An MCP server for a personal notes / knowledge base. Create, search, and retrieve notes — and let an AI assistant (Claude Desktop, Claude Code, …) work with them directly.
The Model Context Protocol is an open standard
that lets AI assistants securely call your tools and read your data. notes-mcp
is a small, self-contained MCP server that turns a folder of notes into something
an assistant can use in natural language — "save this as a note tagged billing",
"what did I write about late check-out?", "summarize everything tagged onboarding".
It exercises all three MCP primitives:
| Primitive | What this server exposes |
|---|---|
| Tools | create_note, update_note, delete_note, get_note, search_notes, list_tags |
| Resources | Every note at note://<id> — the client can pull note content straight into context |
| Prompts | summarize_tag — gathers all notes for a tag and asks the model to synthesize them |
Tech
- TypeScript on Node 20+
@modelcontextprotocol/sdk— the official MCP SDK (stdio transport)- Zod — input schemas / validation
- Zero native dependencies — notes are stored in a plain JSON file, so there's
nothing to compile. The storage layer is isolated in
src/store.ts; swapping in SQLite later would touch only that file. node:test— the test suite spawns the built server and drives it over the real MCP protocol as a client would.
Install & build
pnpm install # or npm install
pnpm build # compiles src → dist
pnpm test # spawns the server and exercises the MCP protocol end-to-end
Where notes are stored
A single JSON file at ~/.notes-mcp/notes.json by default. Override the location
with the NOTES_MCP_DIR environment variable.
Connect it to an assistant
The server speaks MCP over stdio, so any MCP client can launch it.
Claude Desktop — add to claude_desktop_config.json
(~/Library/Application Support/Claude/ on macOS):
{
"mcpServers": {
"notes": {
"command": "node",
"args": ["/absolute/path/to/notes-mcp/dist/index.js"]
}
}
}
Claude Code — one command:
claude mcp add notes -- node /absolute/path/to/notes-mcp/dist/index.js
Then ask the assistant to create, search, or summarize notes — it will call the
tools above. To point the store somewhere specific, add an env var to the config
("env": { "NOTES_MCP_DIR": "/path/to/notes" }).
Try it without a client
The MCP Inspector is the quickest way to poke at the tools by hand:
npx @modelcontextprotocol/inspector node dist/index.js
Scripts
| Command | What it does |
|---|---|
pnpm build |
Compile TypeScript to dist/ |
pnpm watch |
Recompile on change |
pnpm start |
Run the built server on stdio |
pnpm test |
Spawn the server and run the MCP round-trip |
pnpm typecheck |
Type-check without emitting |
How it's structured
src/
store.ts # file-backed note store: CRUD + search (no MCP knowledge)
index.ts # MCP server: wires the store to tools, resources, and prompts
test/
smoke.test.mjs # connects a real MCP client to the spawned server
Keeping store.ts free of any MCP types means the storage layer is testable on its
own and replaceable without touching the protocol wiring.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。