Where are the planning files stored?

Files like task_plan.md and findings.md are created in your current working project directory so they remain accessible alongside your source code.

Does it work automatically?

Yes, it is designed to auto-activate for complex tasks, but it can also be manually invoked using the /planning-with-files command.

How does this prevent AI context loss?

By offloading 'RAM' (context window) to 'Disk' (markdown files), the AI can re-read its own plans and findings to stay oriented during long sessions.

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

Invoke this skill for any complex task requiring more than five tool calls, multi-step research, or when building new projects from scratch.

What is the 3-Strike Error Protocol?

It is a systematic approach to debugging where the AI must diagnose the fix, try an alternative method, and then perform a broader rethink before escalating to the user.

Manus-Style Task Planning

Name: Manus-Style Task Planning
Author: claudiodearaujo

byclaudiodearaujo

0•

Productividad y Flujo de Trabajo

Implements a persistent, file-based planning system for managing complex multi-step tasks and research projects within Claude Code.

This skill equips Claude with a robust 'working memory on disk' by establishing a structured workflow using persistent Markdown files. By maintaining a task plan, findings log, and progress report within the project directory, it ensures that critical context isn't lost during long sessions or complex operations. It follows the 'Manus' methodology, emphasizing documentation before action, error logging, and systematic retries, making it ideal for software architecture, deep research, and large-scale refactoring where the AI's context window might otherwise become saturated.

Características Principales

010 GitHub stars

02Automated session initialization hooks for complex, multi-step workflows

03Three-Strike Error Protocol to prevent repetitive failures and encourage alternative logic

04Read-before-decide matrix to ensure goal alignment before major project changes

05Persistent 'Working Memory' using task_plan.md, findings.md, and progress.md

06Two-Action Rule for immediate data persistence during web research and file operations

Casos de Uso

01Executing complex multi-step software development and refactoring projects

02Conducting deep technical research that spans multiple web searches and documentation reads

03Managing long-running AI sessions where persistent state tracking is required

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add claudiodearaujo/izacenter planning-with-files

For use in Claude.ai and ChatGPT

Características Principales

010 GitHub stars

02Automated session initialization hooks for complex, multi-step workflows

03Three-Strike Error Protocol to prevent repetitive failures and encourage alternative logic

04Read-before-decide matrix to ensure goal alignment before major project changes

05Persistent 'Working Memory' using task_plan.md, findings.md, and progress.md

06Two-Action Rule for immediate data persistence during web research and file operations

Casos de Uso

01Executing complex multi-step software development and refactoring projects

02Conducting deep technical research that spans multiple web searches and documentation reads

03Managing long-running AI sessions where persistent state tracking is required