Things App MCP
An MCP server for Things 3 on macOS that enables AI assistants to create, read, update, and manage tasks and projects. It utilizes the Things URL scheme for write operations and AppleScript for querying data from the app.
README
Things App MCP
An MCP (Model Context Protocol) server for Things 3 on macOS. Enables AI assistants like Claude to create, read, update, and manage your tasks directly in Things.
Features
Write Operations (Things URL Scheme)
| Tool | Description |
|---|---|
add-todo |
Create a new to-do with title, notes, dates, tags, checklist, project/area assignment |
add-project |
Create a new project with to-dos, notes, dates, tags, area assignment |
update-todo |
Update an existing to-do (requires auth-token) |
update-project |
Update an existing project (requires auth-token) |
show |
Navigate to a list, project, area, tag, or specific to-do |
search |
Open the Things search screen |
add-json |
Create complex structures via the Things JSON command |
Read Operations (AppleScript/JXA)
| Tool | Description |
|---|---|
get-todos |
Get to-dos from a list (Inbox, Today, etc.), project, area, or by tag |
get-todo-by-id |
Get a specific to-do by its ID |
get-projects |
Get all projects |
get-project-by-id |
Get a specific project by its ID |
get-areas |
Get all areas |
get-tags |
Get all tags |
search-todos |
Search to-dos by title/notes content |
get-recent-todos |
Get recently modified to-dos |
Requirements
- macOS (required for AppleScript/JXA and
opencommand) - Things 3 installed
- Node.js >= 18
- Things URL Scheme enabled (Things > Settings > General > Enable Things URLs)
Installation
# Clone and build
git clone <repository-url>
cd things-app-mcp
npm install
npm run build
Or install globally:
npm install -g things-app-mcp
Configuration
Claude Desktop
Add to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"things": {
"command": "npx",
"args": ["-y", "things-app-mcp@latest"]
}
}
}
Cursor
Add to your Cursor MCP settings (.cursor/mcp.json):
{
"mcpServers": {
"things": {
"command": "npx",
"args": ["-y", "things-app-mcp@latest"]
}
}
}
Codex
Add to ~/.codex/config.toml:
[mcp_servers.things]
command = "npx"
args = ["-y", "things-app-mcp@latest"]
startup_timeout_sec = 20
tool_timeout_sec = 120
Gemini CLI
Run the following command to register the MCP server:
gemini mcp add things npx -y things-app-mcp@latest
Auth Token Configuration
To use update-todo and update-project, you need your Things auth-token.
Option 1: Environment Variable (Recommended)
Set the THINGS_AUTH_TOKEN environment variable in your MCP client configuration. This avoids needing to pass the token with every request.
Claude Desktop:
{
"mcpServers": {
"things": {
"command": "npx",
"args": ["-y", "things-app-mcp@latest"],
"env": {
"THINGS_AUTH_TOKEN": "your-token-here"
}
}
}
}
Codex (~/.codex/config.toml):
[mcp_servers.things]
command = "npx"
args = ["-y", "things-app-mcp@latest"]
startup_timeout_sec = 20
tool_timeout_sec = 120
[mcp_servers.things.env]
THINGS_AUTH_TOKEN = "your-token-here"
Gemini CLI: Set the environment variable in your shell configuration or pass it when running:
export THINGS_AUTH_TOKEN="your-token-here"
Option 2: Parameter
If the environment variable is not set, you must pass the token as the authToken parameter when calling update tools:
- Open Things on Mac
- Go to Things > Settings > General > Enable Things URLs > Manage
- Copy your authorization token
- Pass it as the
authTokenparameter when calling update tools
Usage Examples
Adding a To-Do
"Add a to-do called 'Buy groceries' scheduled for today with tags 'Errand'"
The AI will call add-todo with:
{
"title": "Buy groceries",
"when": "today",
"tags": "Errand"
}
Creating a Project with To-Dos
"Create a project called 'Launch Website' in the Work area with to-dos: Design mockups, Build frontend, Deploy"
The AI will call add-project with:
{
"title": "Launch Website",
"area": "Work",
"todos": "Design mockups\nBuild frontend\nDeploy"
}
Complex Project via JSON
"Create a vacation planning project with headings for Travel, Accommodation, and Activities"
The AI will call add-json with structured JSON data containing nested headings and to-dos.
Reading To-Dos
"What's on my Today list?"
The AI will call get-todos with { "list": "Today" } and return the structured data.
Updating a To-Do
"Mark the 'Buy groceries' todo as complete"
The AI will first search/get the to-do to find its ID, then call update-todo with the auth-token.
Things URL Scheme Reference
This MCP server implements the full Things URL Scheme v2:
Date Formats
| Format | Example | Description |
|---|---|---|
| Named | today, tomorrow, evening, anytime, someday |
Built-in schedule options |
| Date | 2026-03-15 |
Specific date |
| Date + Time | 2026-03-15@14:00 |
Date with reminder |
| Natural language | next friday, in 3 days |
English natural language (parsed by Things) |
Built-in List IDs (for show tool)
inbox, today, anytime, upcoming, someday, logbook, tomorrow, deadlines, repeating, all-projects, logged-projects
JSON Command Object Types
| Type | Description |
|---|---|
to-do |
A task with title, notes, when, deadline, tags, checklist-items |
project |
A project with title, notes, items (to-dos and headings) |
heading |
A section heading within a project |
checklist-item |
A checklist item within a to-do |
Architecture
things-app-mcp/
src/
index.ts # MCP server entry point with all tool registrations
things-url.ts # Things URL scheme builder (URL construction)
applescript.ts # AppleScript/JXA executor (read operations)
dist/ # Compiled JavaScript output
package.json
tsconfig.json
How It Works
- Write operations construct
things:///URLs and open them via macOSopencommand. Things processes the URL and creates/updates items accordingly. - Read operations use JXA (JavaScript for Automation) scripts executed via
osascriptto query the Things database directly and return structured JSON data.
Development
# Install dependencies
npm install
# Build
npm run build
# Watch mode
npm run dev
# Run directly
npm start
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。