Skills Introduction | Online Tutorial

Skills is an open format for adding professional skills and knowledge to AI agents.

With Skills, you can encapsulate domain-specific expertise, workflows, and best practices into reusable skill packages that AI agents can automatically invoke when needed.

Before learning about Skills, let's first clarify a more fundamental question: What is an AI agent?

You've definitely used tools like ChatGPT, Doubao, and Tongyi Qianwen, right? You type a question, and it gives you a text response. This is what we commonly call a large language model chatbot.

Tools like ChatGPT, Doubao, and Tongyi Qianwen are smart, but they have a fatal flaw: they can only talk, not act.

You ask it: What's the weather like in Beijing today? It can only guess an answer based on what it learned during training, rather than actually checking the weather.

Agent = An AI assistant that can understand what you say, think for itself, and actually get things done.

Chatbots are mouthpieces; agents are true hand replacements.

---

So what is an agent?

Agent = Large language model + ability to take action.

Back to the previous example:

	Chatbot	Agent
You ask: What's the weather like in Beijing today?	Beijing is probably sunny today, around 25°C (guessing)	Actually calls the weather API and tells you: "Today in Beijing it's cloudy, 23°C, 45% humidity" (real data)
You ask: Help me check last month's sales	Sorry, I cannot access your database	Actually queries the database and pulls the data for you
You say: Help me book a high-speed train ticket to Shanghai for tomorrow	You can go to the 12306 website to book	Actually opens 12306 for you, searches for tickets, selects seats, and places the order

The difference in one sentence: Chatbots just talk, agents actually do.

How do agents do it?

An agent roughly consists of the following parts:

Three steps:

1. Understand what you want (powered by large language model)
2. Decide which skill to use (powered by large language model)
3. Execute the skill and deliver results to you (powered by Skills)

Agents you can already see in your life today

In fact, many products are already using agents:

• Siri / Xiao Ai / Tmall Genie — You say "set an alarm," it actually sets it for you, instead of telling you "you can set it yourself"
• GitHub Copilot — Not just chatting, it can actually help you write code and modify code
• Various AI customer service — Not just answering questions, but also helping you check orders, process refunds, change addresses
• Autonomous driving — Also a type of agent, it can perceive the environment, make decisions, and execute operations

---

What are Skills for agents?

Agent's large language model = Brain (responsible for understanding, decision-making, speaking)
Skills = Hands and feet (responsible for doing the actual work)
An agent without Skills is like a person with only a brain but no limbs—knows everything but can do nothing.

The agent itself is very smart, but it needs someone to tell it—in this specific environment, to complete this specific task, which tools should be used, what steps should be followed, and what pitfalls should be avoided.

This instruction manual is called a Skill.

Understanding Skills with a more relatable analogy:

Imagine the agent as a smart new employee with strong abilities, but on the first day at the company knows none of the processes. A Skill is the company's SOP manual (Standard Operating Procedure). With this manual, he knows: how to handle customer complaints, what the reimbursement process is for business trips, what reviews are needed for code submission... No need to ask you every time, he can complete tasks independently.

Technically, Skills is a Markdown file (SKILL.md) containing complete execution guidelines for the task. It is written by humans, then automatically read and followed by the agent when executing tasks.

---

What problems can Skills solve?

Scenario 1: Without Skills

Suppose you ask Claude to generate a Word document for you:

You: Help me generate a Word document for a project proposal
Claude: Okay, here is the project proposal content: (outputs a bunch of Markdown text...)
Result: Claude outputs plain text, not an actual .docx file. Because it doesn't know what libraries are available in your environment, where files should be placed, or what the formatting requirements are.

Scenario 2: With Skills

When the system is equipped with the docx Skill, after receiving the same request, Claude will:

1. Automatically recognize "this is a Word document task" → trigger docx Skill
2. Read the Skill file to understand the correct library (python-docx) and output path (/mnt/user-data/outputs/)
3. Follow the Skill's steps to generate an actual .docx file
4. Hand the file to you for download

Completely different results—one is a chat response, the other is a Word file you can open immediately.

---

Core Value Summary of Skills

After completing this chapter, you need to remember these three points:

• 1. Skills transform agents from chatters to doers: Without Skills, an agent is just a knowledgeable chatbot. With Skills, it becomes an executor that can actually complete tasks.
• 2. Skills are reusable experience packages: Once someone writes the experience of generating Word documents into a Skill, everyone and every call can get stable and consistent results. It won't fail just because someone else asks the question differently.
• 3. Skills are the bridge between humans and AI

Humans write domain knowledge, process experience, and environmental constraints into Skills; AI reads and executes them. Skills are this bridge—humans focus on defining how to do it, AI focuses on actually doing it.

Concept	One-sentence summary
AI Agent	An AI system that can perceive, reason, and act—not just chat, but also execute tasks
Skills	An instruction manual (Markdown file) that tells the agent "how to complete a certain type of task"
Value of Skills	Makes task execution accurate, stable, and reusable

---

Difference Between Skills and MCP

Skills are for knowledge reuse, MCP is for capability extension.

Skills

Knowledge Reuse

• Knowledge sharing: experience, best practices, workflows
• Based on simple Markdown files, anyone can create
• Progressive loading, efficient Token usage
• No server or backend setup required
• Applicable to Web / Desktop / CLI

MCP

Capability Extension

• Function extension: connecting APIs, databases, external tools
• Requires coding ability and server-side configuration
• Loads all tool definitions at startup
• Strong capability for external system integration
• Higher Token consumption and complexity

---

Use Cases for Skills

Agent Skills are applicable to various scenarios, including but not limited to:

• Domain expertise: Encapsulating knowledge from professional fields such as legal review, data analysis, and code standards into skills
• Workflows: Solidifying multi-step task processes into repeatable skills
• Company standards: Encapsulating company-specific code standards, submission processes, document templates, etc. into skills
• Tool integration: Encapsulating usage methods and API calls for specific tools

---

Tools Supporting Agent Skills

Agent Skills is an open standard supported by various AI tools, including:

• Claude Code
• OpenCode
• GitHub Copilot
• OpenAI Codex
• And many other AI agent tools

This means you only need to write a skill once, and you can use it across different tools.