@4fu/pi-todo

This Skill addresses the issue of being unable to persistently track task progress within AI coding sessions. In real-world development, developers often need to manage to-dos across multiple conversation branches, but standard to-do tools either modify the original text or lose context, leading to progress confusion.

Usage is straightforward: you simply use a todo tool with five actions: create to make a task (supporting multi-level checklists), list to view pending tasks or details of a single task, add to append progress updates, complete to mark a task as done, and cancel to cancel a task. You can then manage tasks using natural language instructions. All data is automatically saved within the current conversation branch, and progress automatically follows when you switch branches.

It's ideal for teams or individuals who frequently switch development branches and manage multiple parallel tasks. This is especially useful for users who already employ pi as an AI coding agent, allowing them to track task status without leaving the conversation and avoid context loss.

We recommend using this tool to create task checklists when starting new feature development or project iterations. Note that it primarily relies on pi's conversation branch mechanism; task data is not written to project files or global storage, making it unsuitable for scenarios requiring long-term persistence across separate sessions.