祝融说。

You open your AI tool, wanting to do something—but you don't know which method to use. You'll run into this situation eventually. This chapter helps you find the answer quickly.

You're working on a project and hit a problem: you've implemented the user management feature, but you realize multiple places need similar CRUD operations. You want AI to save you some time, but you're unsure which approach to use—should you let AI complete it automatically, guide it manually, or design the architecture first?

This confusion is normal. AI coding has 14 common skills, each solving a specific type of problem. But you don't need to master all of them—as a beginner, you just need to understand how they're categorized, then focus on the five core skills.

5.1 The Logic Behind the 14 Skills

The 14 skills aren't listed at random; they're categorized by "what problem you're facing." This categorization itself helps you locate the right one quickly.

Planning (use before starting a project)—When you're still unsure how to approach a project, use these skills to clarify your thinking. Execution (use during development)—When you know what to do, use these skills to carry it out. Quality (use when ensuring code quality)—When you need to verify that the code is correct, use these skills to hold the line. Specialized (use in special scenarios)—When you encounter special situations (migrating a legacy system, taking over a half-finished project, etc.), use these skills to handle them.

Here's a quick reference table of the 14 skills:

Planning (use before starting a project)

Execution (use during development)

Quality (use when ensuring code quality)

Specialized (use in special scenarios)

5.2 The Five Skills Beginners Use Most

As a beginner, you don't need to master all 14 skills. These five are enough for most scenarios.

1. Process Coach (Coach)

When to use: Every time you're about to start a new feature.

Why this skill matters most to you: As a beginner, what you need most is "the right habits." The Process Coach helps you build them—it walks you through the Six-Step Workflow, reminding you what to do at each step. You don't have to memorize the process; it guides you step by step alongside. Use it enough, and the Six-Step Workflow becomes muscle memory.

How to use:

Start implementing the note-editing feature.
Requirements: Users can modify a note's title and content on the edit page, and return to the list page after saving.

AI will guide you through the Six-Step Workflow—breaking down milestones, confirming the plan, and inspecting the code.

2. Architecture Design (Architect)

When to use: Before starting a project, or when adding a major feature that affects the system structure.

Why this skill matters to you: The most common mistake beginners make is "write the code first, change direction later." Architecture Design forces you to think before you code—data model, API design, directory structure. Five minutes of thinking now can save you five hours of rework later.

How to use:

Help me design the architecture for this project.
The project is a note-taking app with CRUD functionality for notes.
Please produce a CONTEXT.md blueprint covering the data model, API design, and directory structure.

3. Inspector

When to use: After AI finishes writing code, before you commit to git.

Why this skill matters to you: The most common mistake beginners make is "skipping inspection." AI-generated code looks correct on the surface, but it may harbor hidden problems—logic errors, missing edge cases, security vulnerabilities. Inspection isn't a box-ticking exercise; it's your last chance to hold the line on quality.

How to use:

Inspect the code for the current milestone.
Check whether the functionality is complete, along with code quality, edge cases, and security risks.

4. Advisor

When to use: When you're facing a difficult technical choice or decision.

Why this skill matters to you: Beginners often feel lost when choosing technologies—"Should I use SQLite or PostgreSQL?" "Should I use this library or that one?" The Advisor helps you weigh the pros and cons of your options. But remember: AI gives suggestions, you make the decisions.

How to use:

I need to choose between SQLite and PostgreSQL.
The project isn't expected to have a large amount of data (a few thousand notes), but it may scale later.
Help me analyze the pros and cons of each and give me a recommendation.

5. Automated Workflow (Workflow)

When to use: When a feature's requirements are very clear and you want AI to complete it automatically.

Why this skill suits you: Once you're familiar with the Six-Step Workflow and the requirements are clear, you can let Workflow run the entire process automatically—from coding to inspection, all of it. This is the skill that best showcases the efficiency of AI coding.

How to use:

Fully implement the note search feature.
Users type a keyword in the search box, notes are searched by title with fuzzy matching, and the results update in real time.

5.3 Skill Selection Decision Tree

Not sure which skill to use right now? Refer to this decision tree:

Starting a new project?
├─ Yes → Need the full journey from scratch to deployment?
│   ├─ Yes → Full Build Automation (Job)
│   └─ No → Architecture Design (Architect) → Project Orchestration (Orchestrator)
│
├─ Implementing a new feature?
│   ├─ Requirements clear, want full automation → Automated Workflow (Workflow)
│   ├─ Need guidance, learning as you go → Process Coach (Coach)
│   └─ Multiple features to schedule → Project Orchestration (Orchestrator)
│
├─ Checking code quality?
│   ├─ Routine inspection → Inspector
│   └─ Comprehensive testing → Quality Assurance (QA)
│
├─ Stuck on a decision?
│   └─ Advisor
│
├─ Special scenario?
│   ├─ Want to see the result first → High-Fidelity Prototype (POC)
│   ├─ Reference existing code → Code Cloner (Cloner)
│   ├─ Migrating a legacy system → Legacy System Recon (Legacy Recon)
│   └─ Taking over a half-finished project → Project Handover (Next)
│
└─ Requirements are vague?
    └─ Requirements Analysis

5.4 Skills You Don't Need to Worry About as a Beginner

The following skills are worth knowing about at the entry stage, but you don't need to master them:

• Full Build Automation (Job)—Best for the fully automated journey from scratch to deployment. Beginners should run through the flow manually a few times before trying it.
• Project Orchestration (Orchestrator)—Best for managing multiple features developed in parallel. Beginners should practice with a single feature first.
• Legacy System Recon (Legacy Recon)—Involves legacy systems; a fairly specialized scenario.
• Code Cloner (Cloner)—Requires understanding existing code patterns before you can use it well.
• Project Handover (Next)—Used when taking over someone else's project.

Chapter Summary

The 14 skills are divided into four categories by "what problem you're facing"—Planning, Execution, Quality, and Specialized. Beginners don't need to master them all. Focus on the five core skills: Process Coach (build habits), Architecture Design (avoid wrong directions), Inspector (hold the line on quality), Advisor (support decisions), and Automated Workflow (maximize efficiency). When you're unsure which skill to use, consult the decision tree. In the next chapter, we'll look at the pitfalls beginners are most likely to stumble into.