LABSCO SUMMARY
Storage is a directory you choose, which is the whole privacy story.
There is no service behind this — tasks live wherever MCP_TODO_CACHE_DIR points, so backing them up means backing up that folder, and moving machines means copying it. Small surface, no credentials, nothing leaves the laptop.
What it is
A task manager exposed over MCP, storing its data in a cache directory on your own machine. The assistant becomes the interface: you say what needs doing and it records it.
What you get
- Add new tasks to the list
- Mark tasks complete, edit their content, and delete them
- Sort tasks into categories, so a long list stays findable
Requirements
The project mcp-todo, version 0.1.0, run from a checkout: rye sync to install, then have the client launch src/run.py with the virtualenv's Python. Point MCP_TODO_CACHE_DIR at a directory for the stored data. No account and no key.
Setup effort
Build from source — git clone https://github.com/pushyzheng/mcp-todo
USE CASES
01
Capturing a task mid-conversation
Add it to the list as it comes up, instead of switching to another app and losing the thread.
02
Closing out what you finished today
Mark items complete, or edit the ones whose scope changed.
03
Keeping work and personal items apart
File tasks into categories so each list stays readable.
BEFORE YOU INSTALL
What to know first
Requires
rye on PATH
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
This one is built from source rather than installed with a single command.
Server setup
TERMINAL
git clone https://github.com/pushyzheng/mcp-todo
cd mcp-todo
rye syncMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"mcp-todo": {
"command": "<YOUR_PATH_TO_VENV>/bin/python3",
"args": [
"<YOUR_PATH_TO_SRC>/run.py"
],
"env": {
"MCP_TODO_CACHE_DIR": "<YOUR_MCP_TODO_CACHE_DIR>"
}
}
}
}