Lecture 4: Your First Skill: Hello World Practice

Create your first AI Skill hands-on, completing the leap from theory to practice.

1. Preparation

1.1 Register a Coze Account

  1. Visit https://www.coze.cn
  2. Register with phone number or Douyin account
  3. Complete real-name authentication (if publishing to store)

1.2 Understand Coze Interface

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
┌─────────────────────────────────────────────────────────┐
│  Coze Workspace                                             │
├──────────────┬──────────────────────────────────────────┤
│              │                                              │
│   Sidebar    │              Edit Area                      │
│   ├── Team Space │          ┌─────────────────────┐       │
│   ├── Personal Space │      │   Persona & Response Logic │       │
│   ├── Plugins     │          │   ├── Role Setting    │       │
│   ├── Workflows   │          │   ├── Skill Config   │       │
│   └── Knowledge Base │       │   └── Opening Message │       │
│              │          └─────────────────────┘       │
│              │          ┌─────────────────────┐       │
│              │          │   Preview & Debug Area │       │
│              │          │   └── Right chat window │       │
│              │          └─────────────────────┘       │
└──────────────┴──────────────────────────────────────────┘

2. Create Your First Skill: Weather Assistant

2.1 Create Bot

  1. Click "Create Bot"
  2. Fill in basic information:
    • Name: Weather Assistant
    • Description: Helps you check weather for cities across China
    • Icon: Upload or select system icon
  3. Click "Confirm"

2.2 Configure Persona & Response Logic

In the "Persona & Response Logic" area, enter:

1
2
3
4
5
6
7
8
9
10
11
12
# Role
You are a professional weather query assistant, able to accurately answer various weather-related questions.

## Skills
- Query real-time weather for specified cities
- Query 3-day weather forecast
- Provide clothing and travel advice

## Constraints
- Can only query weather for Chinese cities
- If user doesn't specify city, ask for their location
- Reply in Chinese, friendly and warm tone

2.3 Add Weather Query Plugin

  1. Click "+" on the right side of "Plugins"
  2. Search "weather"
  3. Select "Moji Weather" or "Seniverse Weather" plugin
  4. Click "Add"

2.4 Test Your Skill

In the right preview area, enter:

1
How's the weather in Beijing today?

Expected output:

1
2
Today's weather in Beijing is sunny, temperature 15-25°C, air quality is good, suitable for outdoor activities.
Suggested clothing: Light jacket + long sleeves

3. Advanced: Make Your Skill Smarter

3.1 Add Workflow

Create a new workflow "weather_workflow":

1
2
3
4
5
6
7
8
9
10
11
Start node
    ↓
Parameter extraction node (extract city name)
    ↓
Weather query plugin node (call weather API)
    ↓
Data processing node (format weather information)
    ↓
Suggestion generation node (generate clothing/travel suggestions)
    ↓
End node (return result)

3.2 Workflow Detailed Configuration

Parameter extraction node:

1
2
3
4
5
6
7
8
9
10
11
# Use large model to extract parameters
prompt = """
Extract city name from user input:
User input: {{input}}

Return JSON format:
{
  "city": "City name",
  "date": "Date (today/tomorrow/day after tomorrow)"
}
"""

Weather query node:

1
2
3
4
5
6
# Call Moji weather plugin
plugin: weather_moji
action: get_weather
params:
  city: {{city}}
  days: 3

Suggestion generation node:

1
2
3
4
5
6
7
8
9
10
11
12
prompt = """
Based on the following weather information, generate clothing and travel suggestions:
City: {{city}}
Weather: {{weather}}
Temperature: {{temp_low}}°C - {{temp_high}}°C
Air quality: {{aqi}}

Please provide:
1. One-sentence weather summary
2. Clothing suggestions
3. Travel suggestions
"""

3.3 Bind Workflow to Bot

  1. Return to Bot edit page
  2. Add "weather_workflow" in the "Workflow" area
  3. Add in "Persona & Response Logic":
1
When user asks about weather, call weather_workflow workflow to get weather information.

4. Code Implementation: Build Weather Skill from Scratch

If you want to implement with code, here's a complete example:

4.1 Project Structure

1
2
3
4
5
6
weather-skill/
├── main.py          # Main program
├── intent.py        # Intent recognition
├── weather_api.py   # Weather API wrapper
├── advisor.py       # Suggestion generation
└── requirements.txt # Dependencies

4.2 Core Code

main.py

import json
from intent import IntentClassifier
from weather_api import WeatherAPI


## 🎓 AI 编程实战课程

想系统学习 AI 编程？程序员晚枫的 **AI 编程实战课** 帮你从零上手！

- 👉 **课程报名**：[点击这里报名，前3讲免费试听](https://r7up9.xetslk.com/s/1uP5YW)
- 👉 **免费试看**：[B站免费试看前3讲，先看看适不适合自己](https://www.bilibili.com/cheese/play/ss982042944)