🧠 Claude Plan Action Skill

Stop guessing. Start planning. — A structured planning framework for Claude Code

🌏 Language / 语言：🇨🇳 中文 | 🇬🇧 English

---

A structured task planning methodology packaged as a Claude Code skill. Transforms how you interact with AI for complex coding tasks — eliminating guesswork, reducing rework, and delivering production-quality code on the first try.

📌 Why This?

Before (Direct Prompting)	After (Structured Planning)
🤯 AI guesses your intent → writes wrong code → rework cycle	🎯 AI analyzes requirements → produces a plan → you approve → executes correctly
🧠 Forgets early decisions after 10 conversation rounds	📋 Locks decisions in a written plan before coding starts
🔍 Discovers all problems at the very end	✅ Catches issues at each milestone, one at a time
⏱️ Wastes tokens on hallucinated approaches	💰 Every token works toward an approved direction

Claude Plan Action Skill transforms how you interact with Claude Code for complex coding tasks — eliminating guesswork, reducing rework, and delivering production-quality code on the first try.

🏗️ Workflow

flowchart TB
    subgraph Input["📥 Input"]
        REQ[User Request]
    end
    subgraph Analyze["🔍 1. Analyze"]
        CLS[Task Classification<br/>S/A/B/C Levels]
        DEC[Requirement Breakdown<br/>Scope & Constraints]
    end
    subgraph Plan["📋 2. Plan"]
        AUD[Resource Audit<br/>Agent · Skill · MCP]
        FEA[Feasibility Assessment<br/>5-Dimension Evaluation]
        MIL[Milestone Plan<br/>Nodes · Deliverables]
    end
    subgraph Execute["⚡ 3. Execute"]
        APP[Approval → Execution<br/>Code · Review · Test]
    end

    REQ --> CLS --> DEC
    DEC --> AUD --> FEA --> MIL
    MIL --> APP

    style REQ fill:#6366f1,color:#fff,stroke:none
    style CLS fill:#0ea5e9,color:#fff,stroke:none
    style DEC fill:#0ea5e9,color:#fff,stroke:none
    style AUD fill:#0ea5e9,color:#fff,stroke:none
    style FEA fill:#0ea5e9,color:#fff,stroke:none
    style MIL fill:#0ea5e9,color:#fff,stroke:none
    style APP fill:#10b981,color:#fff,stroke:none

Loading

✨ Core Features

• 🎯 5-Module Planning Framework — Goal breakdown, resource audit, feasibility assessment, milestone plan, task orchestration
• 📊 Task Classification — S/A/B/C levels with appropriate planning depth for each
• ✅ Human-in-the-Loop — No code is written before you approve the plan
• 🔧 Self-Contained — Copy the SKILL.md, register it, and it works immediately

📦 Requirements

Requirement	Details
Claude Code	Latest version
Installation	Copy SKILL.md to .claude/skills/ or use /plan-action

📁 Structure

claude-plan-action-skill/
├── SKILL.md                    # Skill definition (copy to .claude/skills/)
├── skill-files/
│   ├── SKILL.md                # Full skill source
│   ├── PLAN_TEMPLATE.md        # Execution plan template
│   └── AGENT_REGISTRY.example.json
├── docs/
│   ├── plan-action-guide.md    # Comprehensive usage guide
│   └── scan-results.md
├── CHANGELOG.md
├── LICENSE                     # MIT
└── README.md / README.zh.md

❓ FAQ

Does this work with any Claude Code version? Yes. It works with the latest Claude Code CLI. Copy the SKILL.md to your .claude/skills/ directory and it's ready to use.

How does task classification work? Requests are classified as S/A/B/C based on complexity, risk, and number of files involved. Each level has an appropriate planning depth — from full 5-module plan (S/A) to simple checklist (B/C).

Can I modify the planning template? Yes. The PLAN_TEMPLATE.md in skill-files/ is fully customizable. Adjust the sections to match your team's workflow.

Does this work with other AI coding tools besides Claude Code? The skill format is designed for Claude Code, but the planning methodology (5-module framework) is tool-agnostic and can be adapted to any AI coding assistant.

📄 License

MIT © 2026 Ryan Dong

🌟 Star History

👤 About the Author

Ryan Dong — AI Product Manager & Full-Stack Developer

I bridge the gap between AI capabilities and production-ready software. My work spans the full stack: from designing AI-powered product features and integrating LLM APIs, to building modular backend services and shipping clean, documented code.

Role	Focus
🧠 AI Product Manager	Product strategy, AI feature design, prompt engineering, model selection
💻 Full-Stack Developer	Python, FastAPI, Google Apps Script, automation pipelines, API integration

This repository is part of a personal toolbox — a growing collection of practical, reusable modules that solve real automation problems. Each project is designed to be independently useful and easily integrated into larger systems.

📬 donglinfei@gmail.com — open to business discussions, collaborations, and recruiting inquiries.

📬 Contact

Ryan Dong — donglinfei@gmail.com