ThingsSAFE
Enhanced FastMCP implementation of the Things MCP server for Claude and Windsurf
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
This Model Context Protocol (MCP) server lets you use Claude Desktop to interact with your task management data in Things app. You can ask Claude to create tasks, analyze projects, help manage priorities, and more.
This server leverages the Things.py library and the Things URL Scheme, with additional reliability features including:
- Robust error handling with exponential backoff and retry mechanisms
- Circuit breaker pattern to prevent cascading failures
- Dead letter queue for failed operations
- Intelligent caching for improved performance
- Comprehensive logging with structured JSON output
- AppleScript bridge for operations that fail with URL schemes
- Rate limiting to prevent overwhelming the Things app
- Extensive test suite for reliability
Why Things MCP?
This MCP server unlocks the power of AI for your task management:
- Natural Language Task Creation: Ask Claude to create tasks with all details in natural language
- Smart Task Analysis: Get insights into your projects and productivity patterns
- GTD & Productivity Workflows: Let Claude help you implement productivity systems
- Seamless Integration: Works directly with your existing Things 3 data
Features
- Access to all major Things lists (Inbox, Today, Upcoming, etc.)
- Project and area management
- Tag operations
- Advanced search capabilities
- Recent items tracking
- Detailed item information including checklists
- Support for nested data (projects within areas, todos within projects)
Installation Options
There are multiple ways to install and use the Things MCP server:
Option 1: Install from PyPI (Recommended)
Prerequisites
- Python 3.12+
- Claude Desktop
- Things 3 ("Enable Things URLs" must be turned on in Settings -> General)
*
3bb14a30f449OBSERVED · 2026-10-08Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add things3-enhanced-mcp --env THINGS_AUTH_TOKEN=${THINGS_AUTH_TOKEN} -- uvx things3-enhanced-mcp{
"mcpServers": {
"things3-enhanced-mcp": {
"command": "uvx",
"args": [
"things3-enhanced-mcp"
],
"env": {
"THINGS_AUTH_TOKEN": "${THINGS_AUTH_TOKEN}"
}
}
}
}Exposed tools (22)
18 read · 4 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
add-project | write | |
add-todo | write | |
get-anytime | read | Get todos from Anytime list |
get-areas | read | |
get-cache-stats | read | Get cache performance statistics |
get-inbox | read | Get todos from Inbox |
get-logbook | read | |
get-projects | read | |
get-recent | read | |
get-someday | read | Get todos from Someday list |
get-tagged-items | read | |
get-tags | read | |
get-today | read | Get todos due today |
get-todos | read | |
get-trash | read | Get trashed todos |
get-upcoming | read | Get upcoming todos |
search-advanced | read | |
search-items | read | |
search-todos | read | |
show-item | read | |
update-project | write | |
update-todo | write |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (3)
return hashlib.md5(key_string.encode()).hexdigest()
if k.startswith(hashlib.md5(f"{operation}:".encode()).hexdigest()[:8])]curl -LsSf https://astral.sh/uv/install.sh | sh
Gates applied: no_behavioural_pass.
3bb14a30f449full audit observations/trust-audit/mcp-server/excelsier__things-1.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | 3bb14a30f449 | SAFE | B | 89 | first audit |
Questions
What is the Things MCP server?
Enhanced FastMCP implementation of the Things MCP server for Claude and Windsurf
What tools does Things expose?
22 in total: 18 read-only, 4 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is Things safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does Things need?
It reads THINGS_AUTH_TOKEN from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does Things run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as things3-enhanced-mcp.
How current is this page?
The grade is for one exact copy of the source (3bb14a30f449), read on 2026-10-08. The repository is watched and re-audited when it changes.