things3-mcp-server
Enables AI agents to read and write to-dos, lists, areas, and projects in Things 3 on macOS via AppleScript, allowing task management workflows like triaging, logging, and rescheduling.
README
things3-mcp-server
MCP server for Things 3 on macOS. Read and write to-dos, lists, areas and projects — so an agent can drive the same task workflows a human does in Things (triage Today, log completed work, file new tasks, reschedule).
Built for the KI-OS workflow: it replaces the ad-hoc AppleScript snippets used
for /review, the purchase-decision framework and the loop↔Things sync with a
proper tool surface.
How it works
All access goes through AppleScript (osascript), invoked with an
on run argv handler so user-provided values (task titles, notes, tags) are
passed as real arguments — never interpolated into the script, so there is no
injection risk. No external Python dependency beyond the MCP SDK; nothing reads
the Things SQLite database directly, which keeps writes safe and avoids schema
coupling.
Requires the Things 3 app to be installed and running. On first use macOS will ask to grant Automation permission to control Things 3 — approve it for the host process (Claude / your terminal).
Tools
| Tool | Kind | Purpose |
|---|---|---|
list_todos(scope_type, scope_name, status) |
read | To-dos in a list / area / project |
get_today() |
read | Everything in the Today list |
get_inbox() |
read | Everything in the Inbox |
get_completed(days) |
read | Recently completed to-dos (Logbook) |
search_todos(query, include_completed) |
read | Find to-dos by name substring |
get_todo(todo_id) |
read | A single to-do by id |
list_areas() |
read | All areas |
list_projects(area) |
read | Projects (optionally within an area) |
list_tags() |
read | All tag names |
create_todo(title, notes, area, project, tags, when, deadline) |
write | Create a to-do |
complete_todo(query, todo_id, exact) |
write | Complete open to-do(s) by name or id |
cancel_todo(query, todo_id, exact) |
write | Cancel open to-do(s) — reversible, no delete |
update_todo(query, todo_id, tags, when, deadline, move_to_list, exact) |
write | Update the first matching open to-do |
add_tags(tags, query, todo_id, exact) |
write | Append tags without dropping existing ones |
Write tools accept either todo_id (exact, safest — e.g. an id from
create_todo) or a name query (exact=True for an exact match instead of a
substring).
Todo objects carry: id, name, status, tags, due (deadline, ISO),
when (scheduled/activation date, ISO), project, area, completion (ISO),
notes. Dates are YYYY-MM-DD strings; when accepts today, tomorrow,
anytime, someday or an ISO date.
Install
cd ~/workspace/things3-mcp-server
python3 -m venv .venv
.venv/bin/pip install -e .
Register with Claude Code (user scope)
claude mcp add things3 --scope user -- \
~/workspace/things3-mcp-server/.venv/bin/things3-mcp-server
Design note
The MCP stays a generic Things wrapper. Domain semantics — e.g. the ADHS
48h purchase-decision rule (48h-Liste tag, +2 days, then Spontankauf +7) —
live in the KI-OS skills, not here. update_todo(query, tags=..., move_to_list="Anytime")
is exactly the primitive the 48h→Spontankauf transition needs.
Scope / non-goals
- The interactive "Einkäufe ADHS" Shortcut is deliberately not wrapped (it needs UI dialogs and stays manual).
- No delete tool by design —
cancel_todo(reversible) covers "drop this task" (e.g. a rejected purchase decision); hard deletion is intentionally left out.
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.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
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.