Notebook MCP Server
Enables to create notebooks and notes with tags, and perform full-text search across notes using FTS5 syntax, all stored locally in SQLite.
README
Notebook MCP Server
A simple Model Context Protocol server for creating notebooks and notes with tags and full-text search. Notes are stored in a single SQLite database with an FTS5 index for search — no external services required.
Tools
| Tool | Description |
|---|---|
create_notebook |
Create a notebook (name, optional description). |
create_note |
Add a note to a notebook (notebook, title, content, optional tags). |
read_note |
Read a full note by note_id. |
update_note |
Update a note's title, content, and/or tags (only provided fields change). |
search_notes |
Full-text search over titles, content, and tags; filter by notebook and/or tag. Supports FTS5 syntax (AND, OR, NOT, "exact phrase", prefix*); queries that aren't valid FTS5 syntax (e.g. e-mail) are matched as literal words instead. |
Errors (duplicate notebook, missing note, invalid arguments) are raised as
exceptions, so MCP clients receive them as proper tool errors (isError), not
as text that looks like a successful result.
Quick start (no clone needed)
The only prerequisite is uv.
The commands below are copy-pasteable as-is — no paths to edit.
Claude Code:
claude mcp add notebook -- uvx --from git+https://github.com/lucidbard/notebook-mcp-server notebook-mcp
Claude Desktop — add to claude_desktop_config.json
(Settings → Developer → Edit Config), then restart the app:
{
"mcpServers": {
"notebook": {
"command": "uvx",
"args": ["--from", "git+https://github.com/lucidbard/notebook-mcp-server", "notebook-mcp"]
}
}
}
Try it
Ask your client something like:
Create a notebook called "ideas", then save a note in it titled "Search UX" tagged
uxandsearch, with a few thoughts on query autocomplete.
Then later:
Search my notes for anything about autocomplete.
The model will call create_notebook, create_note, and search_notes for you.
Where your notes live
All data is one SQLite file at ~/.notebook-mcp/notebook.db (created on first
run). Back up that file to back up your notes; delete it to start fresh. Set
the NOTEBOOK_DB environment variable to use a different location — handy for
keeping separate databases per project.
Installing from a clone
If you'd rather not run from GitHub directly (Python 3.10+ required):
git clone https://github.com/lucidbard/notebook-mcp-server.git
cd notebook-mcp-server
pip install .
claude mcp add notebook -- notebook-mcp
pip install . puts a notebook-mcp command on your PATH, so the client
config never needs a file path. (If you do point a client at the script
directly, use the absolute path to notebook_server.py — placeholder
paths like /path/to/... fail to connect, and Git Bash silently rewrites
them into C:/Program Files/Git/... on Windows.)
Development
pip install -r requirements.txt
python notebook_server.py # run the server (MCP over stdio)
python test_fixes.py # tool, error-path, and search tests
python stdio_test.py # end-to-end test over MCP stdio
python stdio_test.py uvx --from . notebook-mcp # same, against the packaged install
Both test scripts use a throwaway temporary database — they never touch
~/.notebook-mcp/notebook.db.
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。