When should I use the planning-with-files skill?

Use it for any non-trivial task involving three or more steps, research, or complex building; skip it for simple questions or quick edits.

How does this improve Claude Code's performance?

By storing goals and progress in physical files, it prevents Claude from 'forgetting' the original objective during long, multi-step conversations.

What is the 3-file pattern used in this skill?

It is a workflow that creates task_plan.md for progress, notes.md for research findings, and a final deliverable file to keep the context organized.

Does it automatically update my project status?

Yes, the skill follows a 'Update After Act' rule, ensuring the task_plan.md is refreshed after every major phase or error encountered.

Manus-Style Task Planning

Name: Manus-Style Task Planning
Author: kjgarza

bykjgarza

生产力与工作流

Implements a structured three-file workflow to maintain persistent project memory and track progress through complex tasks.

关于

This skill transforms Claude's workflow by adopting the 'Manus' philosophy of using persistent markdown files as external working memory. By mandating a three-file structure—task_plan.md for milestones, notes.md for research, and a dedicated deliverable file—it prevents context drift and ensures that goals remain visible throughout long-running projects. It is particularly effective for multi-step engineering tasks, deep research, and complex debugging where logging decisions and errors is critical for maintaining project state across multiple tool calls.

主要功能

Context optimization by offloading large data to disk
Standardized 3-file project architecture (Plan, Notes, Deliverable)
0 GitHub stars
Persistent phase-based progress tracking with checkboxes
Mandatory error logging and decision documentation
Manus-style 'read-before-decide' attention refresh loops

使用场景

Tracking complex debugging sessions by logging every error and resolution attempt
Managing multi-step software builds that require state persistence
Conducting deep research where findings must be synthesized before execution

关于

主要功能

Context optimization by offloading large data to disk
Standardized 3-file project architecture (Plan, Notes, Deliverable)
0 GitHub stars
Persistent phase-based progress tracking with checkboxes
Mandatory error logging and decision documentation
Manus-style 'read-before-decide' attention refresh loops

使用场景

Tracking complex debugging sessions by logging every error and resolution attempt
Managing multi-step software builds that require state persistence
Conducting deep research where findings must be synthesized before execution