This guide explains how to upgrade from the legacy shell script (v2.x) to the new Python-based version (v3.0).
Version 3.0 is a complete rewrite in Python, offering:
- MCP Integration: AI agents interact directly with your task list
- Dual Interfaces: MCP Server + Standard CLI
- Cross-Platform: Native support for macOS, Linux, and Windows
- Zero-Data Migration: Fully compatible with existing
TODO.mdand.ai-todo/
The easiest migration path is to skip CLI installation entirely and use MCP with uvx:
Add this to your project's .cursor/mcp.json:
{
"mcpServers": {
"ai-todo": {
"command": "uvx",
"args": ["ai-todo", "serve", "--root", "${workspaceFolder}"]
}
}
}Prerequisite: Install uv first:
curl -LsSf https://astral.sh/uv/install.sh | shThat's it! Your AI agent can now manage tasks directly.
If you prefer command-line access, install the CLI globally:
# Using uv (recommended)
uv tool install ai-todo
# Or with pipx
pipx install ai-todoPrerequisite: Python 3.10+
curl -LsSf https://astral.sh/uv/install.sh | shmacOS:
brew install pipx
pipx ensurepathLinux/Windows: Follow the pipx installation guide.
Since v3.0 uses the exact same file formats (TODO.md, .ai-todo/config.yaml), no data migration is required.
Note: If you have an existing
.todo.ai/directory, it will be automatically migrated to.ai-todo/on first use.
ai-todo versionIf you want to remove the legacy script:
git rm todo.ai todo.bash # if present| Old (Shell) | New (CLI) |
|---|---|
./todo.ai list |
ai-todo list |
./todo.ai add "Task" |
ai-todo add "Task" |
./todo.ai complete 1 |
ai-todo complete 1 |
Core command syntax is preserved.
See MCP_SETUP.md for detailed Cursor integration instructions.
"Command not found: ai-todo"
Ensure your tool directory is in PATH:
# For uv
echo $PATH | grep -q ".local/bin" || echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
# For pipx
pipx ensurepathRestart your terminal after updating PATH.
"Config/Data Issues"
Run ai-todo list to verify your existing TODO.md parses correctly. The tool preserves all existing data.
- FAQ: Common questions
- MCP Setup: Cursor integration
- Issues: https://github.com/fxstein/ai-todo/issues