spacecadet

An MCP server that gives AI assistants reliable task management through Emacs org-mode.

Why?

LLMs are unreliable at tracking tasks in conversation context. They lose count, forget deadlines, and hallucinate state. Spacecadet solves this by delegating task storage and retrieval to org-mode -- a deterministic, structured system with 100% accuracy.

What it does

Spacecadet exposes 14 MCP tools that let any AI assistant:

• Create, update, delete, and list tasks with priorities, tags, deadlines, and scheduled dates
• Search and query using org-mode's full match syntax
• Track time with clock-in/clock-out and reporting
• Organize with notes, custom properties, and refiling

Every task gets a unique ID for reliable programmatic access.

Quick start

git clone https://github.com/jamesdp3/spacecadet.git
cd spacecadet
./setup.sh

Then configure your MCP client -- see Getting Started for details.

How it works

AI Client (Claude Desktop, Claude Code, etc.)
    |  MCP protocol (stdio)
    v
server.py  (Python, FastMCP)
    |  emacsclient --eval
    v
Emacs daemon  (persistent, started on first request)
    |  org-mode API
    v
~/spacecadet-tasks/*.org  (plain text)

A dedicated Emacs daemon starts on the first request and stays running for fast subsequent calls. Tasks are stored as plain .org text files -- point them at a cloud-synced folder (iCloud, Dropbox, etc.) to access from any machine. No interference with your personal Emacs setup. See Architecture for the full picture.