什么是 nb-mcp-server?
封装 nb CLI 的 MCP server,为 LLM 提供更友好的笔记记录与管理能力。
README
nb-mcp
MCP server wrapping the nb CLI for LLM-friendly note-taking.
Motivation
Using nb directly via shell has two problems for LLM assistants:
-
Backtick escaping: Markdown content with backticks triggers shell command substitution, corrupting notes.
-
Notebook context:
nbassumes a default notebook, making per-project use awkward.
This MCP server solves both by:
- Accepting content as JSON parameters (no shell escaping needed)
- Qualifying all commands with an explicit notebook
Quick Start
Prerequisites
Install nb by following the official instructions:
nb installation guide.
Installation
From crates.io:
cargo install nb-mcp-server
Or download a prebuilt binary from GitHub Releases.
Build from Source
cargo build --release
Run
With default notebook from environment:
NB_MCP_NOTEBOOK=myproject ./target/release/nb-mcp
Or via CLI argument (takes precedence):
./target/release/nb-mcp --notebook myproject
Disable commit and tag signing in the notebook repository:
./target/release/nb-mcp --notebook myproject --no-commit-signing
Print the installed version:
./target/release/nb-mcp --version
Show the resolved notebook path and state directory:
./target/release/nb-mcp --show-paths
MCP Configuration
Add to your MCP client configuration (e.g., .mcp.json):
{
"mcpServers": {
"nb": {
"command": "/path/to/nb-mcp",
"args": ["--notebook", "myproject"]
}
}
}
Commands
All commands are accessed via the nb tool with a command parameter to
reduce the token footprint of the MCP server.
The args field must be a JSON object. Stringified JSON payloads are rejected.
Notes
| Command | Description | Key Arguments |
|---|---|---|
nb.add | Create a note | title, content, tags[], folder |
nb.show | Read a note | id (alias: selector) |
nb.edit | Update a note | id (alias: selector), content, mode (replace default, append, prepend) |
nb.delete | Delete a note | id (alias: selector) |
nb.move | Move or rename a note | id (alias: selector), destination |
nb.list | List notes | folder, tags[], limit ([ ] / [x] indicate todo status; leading glyphs are item markers) |
nb.search | Full-text search | queries[] (required), mode (any default, all), tags[] |
Todos
| Command | Description | Key Arguments |
|---|---|---|
nb.todo | Create a todo | description, optional tasks[], tags[] |
nb.do | Mark complete | id (alias: selector), optional task_number |
nb.undo | Reopen | id (alias: selector), optional task_number |
nb.tasks | List todos | optional status (open or closed), optional recursive (true default) |
Organization
| Command | Description | Key Arguments |
|---|---|---|
nb.bookmark | Save a URL | url, title, tags[], comment |
nb.import | Import file/URL | source, folder, filename, convert |
nb.folders | List folders | parent |
nb.mkdir | Create folder | path |
nb.notebooks | List notebooks only | (none) |
nb.status | Notebook info | (none) |
Examples
Create a note with code:
{
"command": "nb.add",
"args": {
"title": "API Design Notes",
"content": "# API Design\n\nUse `GET /items` for listing.\n\n```python\nresponse = client.get('/items')\n```",
"tags": ["design", "api"],
"folder": "docs"
}
}
Search for notes:
{
"command": "nb.search",
"args": {
"queries": ["API", "design"],
"mode": "any",
"tags": ["design"]
}
}
Tagging Suggestions
For multi-LLM projects, consider using consistent tag prefixes (optional). Example categories and prefixes:
| Category | Pattern | Examples |
|---|---|---|
| Collaborator | llm-<name> | llm-claude, llm-gpt |
| Component | component-<name> | component-api, component-ui |
| Task type | task-<type> | task-bug, task-feature |
| Status | status-<state> | status-review, status-blocked |
Configuration
Notebook Resolution
Priority order:
- Per-command
notebookargument (highest) - CLI
--notebookflag NB_MCP_NOTEBOOKenvironment variable- Git-derived default from the master worktree path
If no notebook can be resolved, commands fail with a configuration error. The
server does not fall back to nb's default notebook.
If the resolved notebook does not exist, the server creates it automatically.
Use --no-create-notebook to disable automatic creation.
Logging
Logs are written to ~/.local/state/nb-mcp/{project}--{worktree}.log (XDG-compliant).
For Git worktrees, logs are named after both the master project and the worktree basename to avoid collisions between multiple MCP server instances.
Use --show-paths to print the resolved notebook path and state directory.
Control log level with RUST_LOG:
RUST_LOG=debug nb-mcp --notebook myproject
Commit Signing
Use --no-commit-signing to disable commit and tag signing in the notebook
repository. The server updates the notebook repository's local Git config so
signing prompts do not block MCP tool calls.
Contributing
See the contribution guide and code of conduct:
License
常见问题
nb-mcp-server 是什么?
封装 nb CLI 的 MCP server,为 LLM 提供更友好的笔记记录与管理能力。
相关 Skills
MCP构建
by anthropics
聚焦高质量 MCP Server 开发,覆盖协议研究、工具设计、错误处理与传输选型,适合用 FastMCP 或 MCP SDK 对接外部 API、封装服务能力。
✎ 想让 LLM 稳定调用外部 API,就用 MCP构建:从 Python 到 Node 都有成熟指引,帮你更快做出高质量 MCP 服务器。
Slack动图
by anthropics
面向Slack的动图制作Skill,内置emoji/消息GIF的尺寸、帧率和色彩约束、校验与优化流程,适合把创意或上传图片快速做成可直接发送的Slack动画。
✎ 帮你快速做出适配 Slack 的动图,内置约束规则和校验工具,少踩上传与播放坑,做表情包和演示都更省心。
MCP服务构建器
by alirezarezvani
从 OpenAPI 一键生成 Python/TypeScript MCP server 脚手架,并校验 tool schema、命名规范与版本兼容性,适合把现有 REST API 快速发布成可生产演进的 MCP 服务。
✎ 帮你快速搭建 MCP 服务与后端 API,脚手架完善、扩展顺手,尤其适合想高效验证服务能力的开发者。
相关 MCP Server
Slack 消息
编辑精选by Anthropic
Slack 是让 AI 助手直接读写你的 Slack 频道和消息的 MCP 服务器。
✎ 这个服务器解决了团队协作中需要 AI 实时获取 Slack 信息的痛点,特别适合开发团队让 Claude 帮忙汇总频道讨论或发送通知。不过,它目前只是参考实现,文档有限,不建议在生产环境直接使用——更适合开发者学习 MCP 如何集成第三方服务。
by netdata
io.github.netdata/mcp-server 是让 AI 助手实时监控服务器指标和日志的 MCP 服务器。
✎ 这个工具解决了运维人员需要手动检查系统状态的痛点,最适合 DevOps 团队让 Claude 自动分析性能数据。不过,它依赖 NetData 的现有部署,如果你没用过这个监控平台,得先花时间配置。
by d4vinci
Scrapling MCP Server 是专为现代网页设计的智能爬虫工具,支持绕过 Cloudflare 等反爬机制。
✎ 这个工具解决了爬取动态网页和反爬网站时的头疼问题,特别适合需要批量采集电商价格或新闻数据的开发者。不过,它依赖外部浏览器引擎,资源消耗较大,不适合轻量级任务。