OthmanAdi/planning-with-files

⭐ 20,746  ·  #1  ·  Python

Claude Code skill implementing Manus-style persistent markdown planning — the workflow pattern behind the $2B acquisition.

Python adal agent-skills antigravity Skill

Project Analysis

🎯 Positioning	Agent capability enhancement
💡 Core Value	Provides standardized Skills and Prompt templates for AI coding Agents, covering specific scenarios (code review, debugging, architecture design, etc.), enabling higher quality output in these scenarios
👥 Suitable For	Developers using Agent tools like Claude Code/Cursor/Codex, aiming to improve Agent performance on specific tasks

Why It Deserves Attention

20,746 Stars, with good community activity, indicating it solves real pain points. Developed in Python.

Transforms Manus-style planning into a reusable Claude Code skill.

Core Features

• Markdown Files as Planning: Uses .md files as planning carriers to achieve AI Agent task decomposition and execution tracking, without complex databases.
• Manus Workflow Replication: Fully replicates the core planning pattern behind the $2B acquisition—persistent, iterable step management.
• Multi-Agent Framework Compatibility: Besides Claude Code, it also adapts to mainstream Agent frameworks like Copilot, Mastra, and OpenClaw, reducing migration costs.
• Lightweight Skill Encapsulation: Exists as a skill, not a standalone application, directly injectable into existing Agent workflows with zero intrusion.
• State Persistence: Planning files support progress marking, task completion status recording, and resumption after interruption.

Technical Architecture

• Tech Stack: Pure Python implementation with minimal dependencies; core logic revolves around file I/O and Markdown parsing.
• Code Structure: skills/ directory categorized by Agent framework (claude/, copilot/, mastra/, etc.), each subdirectory containing independent skill definition files.
• Design Highlights:
  • Adopts a "planning as file" model, moving state management from memory/database to the file system, greatly simplifying deployment complexity.
  • Achieves cross-framework planning definition reuse through a unified Markdown template format.
  • No external dependencies, relying only on the Python standard library, embodying a "less is more" design philosophy.

Quick Start Guide

1. Clone Repository: git clone https://github.com/OthmanAdi/planning-with-files.git
2. Select Framework: Navigate to the corresponding directory, e.g., cd planning-with-files/skills/claude/
3. Inject Skill: Copy the planning.md file to the Agent's working directory
4. Start Agent: Load the skill file in Claude Code to begin using planning functionality
5. Customize Planning: Modify the task list in the .md file; the Agent will automatically recognize and execute it

Strengths, Weaknesses, and Use Cases

Strengths

• Extremely Low Cognitive Load: No need to learn new tools; Markdown is planning, files are state.
• Framework Agnostic: A single planning template adapts to multiple Agent frameworks, reducing repetitive work.
• Lightweight and Auditable: All plans exist as files, enabling version control and manual review.

Weaknesses

• Standalone Limitations: File-level state management is unsuitable for distributed, multi-Agent collaboration scenarios.
• Functional Boundaries: Lacks advanced planning features (e.g., dependency graphs, resource scheduling, dynamic re-planning).
• Ecosystem Dependency: Requires specific Agent frameworks to function; not a standalone solution.

Use Cases

• Individual Developers: Quickly build AI-assisted task management systems.
• Small Teams: Implement traceable planning workflows in Agent tools like Claude Code.
• Prototype Validation: Verify the feasibility of the "planning as file" model in specific business contexts.

Community and Popularity

• Star Count: 20,746, with rapid growth, reflecting strong market demand for Manus pattern reuse.
• Update Activity: Last updated in May 2026; the project is still under continuous iteration.
• Topic Popularity: Covers hot Agent topics like adal, agent-skills, manus, mastra, with high community attention.
• Fork Trend: Active fork count indicates many developers are building upon and customizing this project.

Review: This project precisely targets the pain point of "planning capability" in the AI Agent field, achieving high-value workflow replication with a minimal technical solution. While not a general framework, as a skill component, its design approach and practicality are worth referencing.

Technical Information

• 💻 Language: Python
• 📂 Topics: adal, agent-skills, antigravity, claude, claude-code
• 🕐 Updated: 2026-01-02
• 🔗 Visit GitHub Repository

---

Data updated on 2026-05-09 · Star count based on actual GitHub data