AI-Assisted Programming

RuoYi AI documentation

This guide covers RuoYi AI. Help improve it by reporting problems in an issue.

This tutorial uses Claude Code to illustrate AI-assisted development. The beginner section covers installation, configuration, and daily operations. The advanced section introduces MCP integrations and Skills for specialized tasks.

The examples follow the conventions in this project's Chinese tutorial. Check prompts and available settings in your installed CLI or IDE version; shortcuts, permission behavior, and configuration fields can vary. Original sample prompts and code are retained where they refer to the same demonstration data.

Getting started

Prerequisites

For VS Code, the source tutorial recommends version 1.98.0 or later and Git for Windows on Windows. For JetBrains IDEs, such as IntelliJ IDEA, PyCharm, or WebStorm, it recommends version 2025.1 or later. The npm-based CLI examples below use Node.js 18 or later; follow the actual requirements of the version you install.

Install Claude Code

1. VS Code extension

From the extension marketplace:

Open Extensions with Ctrl+Shift+X.
Search for Claude Code.
Install the official Anthropic extension.

From the command line:

bash

npm install -g @anthropic-ai/claude-code

The source tutorial uses this CLI path to initialize the VS Code integration. Follow any installation prompts shown by your version.

2. JetBrains integration

Install the command-line tool:

bash

npm install -g @anthropic-ai/claude-code

Open the IDE's Settings → Plugins, search for Claude Code, install the official plugin, and restart the IDE. The source tutorial shows the Beta-labeled plugin.
Verify installation:

bash

claude-code --version

1. Initialize project context

Run /init to analyze the project structure, identify its technology stack, and create a CLAUDE.md file describing project conventions. Review that file so it accurately reflects the repository.

2. Switch permission modes

The tutorial uses Shift+Tab to cycle through these modes:

Mode	Behavior	Indicator
Normal	Uses the configured permission checks and prompts	No special indicator
Auto-Accept	Accepts edits without individual edit confirmations	`⏵⏵ accept edits on`
Plan	Analyzes and plans without editing project files	`⏸ plan mode on`

When to use Plan Mode

Use it when a change spans several files, the approach is still undecided, or the work benefits from reviewing the plan before edits.

Choose a mode at startup

bash

claude --permission-mode plan

3. Reference files and directories

Type @ followed by a path, using completion to select the intended file or directory. This supplies focused context for a particular module. Completion behavior can differ between terminals and IDE integrations.

4. Export a conversation

Use /export to save a conversation as Markdown for archiving, reviewing a complex investigation, or sharing with your team. Follow the current terminal's export prompts.

5. Request deeper reasoning

The source tutorial uses an ultrathink: prefix for architecture, difficult debugging, and refactoring tradeoffs. Its effect depends on the installed version and configuration; do not assume the text alone guarantees a particular reasoning budget.

6. Paste screenshots

Paste a screenshot into the supported input area, using Ctrl+V where supported, and describe the issue. For example:

[Screenshot] The button's spacing is incorrect. Please fix it to match the adjacent controls.

7. Correct direction early

Press Esc to interrupt when the assistant takes the wrong direction, then give a specific correction. There is no need to wait for a long response to finish.

8. Navigate previous prompts

Use the Up and Down arrow keys to navigate prior input, repeat related work, or recall a previous command.

9. Run a shell command

Prefix a command with ! to run it and bring the output into the conversation. For example, !npm --version checks the installed npm version.

10. Search prompt history

Use Ctrl+R and a keyword to find a previous prompt, particularly a long command or detailed request.

11. Skip permission checks with care

The source tutorial includes this startup option:

bash

claude --dangerously-skip-permissions

It bypasses permission confirmations and can execute commands that change or delete files. Use it only in an environment where the task and executable content are fully trusted and the consequences are understood.

Appendix: configuration examples

Configuration locations

bash

# Windows
%USERPROFILE%\.claude.json

# macOS/Linux
~/.claude.json

.claude.json

The tutorial's .claude.json example contains basic setup information and MCP server definitions. Internal fields can differ by version; prefer the installed tool's supported configuration flow.

Complete example

json

{
  "hasCompletedOnboarding": true,
  "acceptedTos": true,
  "autoUpdates": false,
  "installMethod": "npm",
  "userID": "00000000-guest-user-bypass-config-template-00000000",
  "firstStartTime": "2025-01-01T00:00:00.000Z",
  "sonnet45MigrationComplete": true,
  "opus45MigrationComplete": true,
  "opusProMigrationComplete": true,
  "thinkingMigrationComplete": true,
  "cachedChromeExtensionInstalled": false,
  "mcpServers": {
    "bing-search": {
      "command": "npx",
      "args": ["-y", "bing-cn-mcp"]
    }
  }
}

Fields in the example

Field	Meaning	Example value
`hasCompletedOnboarding`	Whether onboarding is complete	`true`
`acceptedTos`	Terms acceptance state	`true`
`autoUpdates`	Automatic updates	`false`
`installMethod`	Installation method	`npm`
`userID`	User identifier	Generated by the tool
`mcpServers`	MCP server definitions	`{}`

settings.json

The project settings example configures environment variables and project-specific behavior.

Location

bash

# 项目根目录
./settings.json

Complete example

json

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.anthropic.com",
    "ANTHROPIC_AUTH_TOKEN": "sk-ant-your-token-here",
    "ANTHROPIC_MODEL": "claude-opus-4-6",
    "ANTHROPIC_SMALL_FAST_MODEL": "claude-haiku-4-5-20251001",
    "NODE_ENV": "development"
  }
}

Environment fields

Field	Meaning	Required by this example	Example
`ANTHROPIC_BASE_URL`	API base URL	No	`https://api.anthropic.com`
`ANTHROPIC_AUTH_TOKEN`	API authentication token	Yes	`sk-ant-xxxxx`
`ANTHROPIC_MODEL`	Default model	No	`claude-opus-4-6`
`ANTHROPIC_SMALL_FAST_MODEL`	Small, fast model	No	`claude-haiku-4-5-20251001`
`NODE_ENV`	Runtime environment	No	`development`

Configuration steps

Step 1: create .claude.json

bash

# 1. 进入用户主目录
cd ~

# 2. 创建 .claude.json 文件（如果不存在）
# 复制上面的完整配置示例

# 3. 替换其中的 sk-xxx 为你的实际 API Key

Step 2: optionally create settings.json

bash

# 1. 进入项目根目录
cd /path/to/your/project

# 2. 创建 settings.json 文件
# 复制上面的完整配置示例

# 3. 配置项目特定的环境变量

Step 3: verify configuration

bash

# 查看配置是否正确
cat ~/.claude.json

# 在 Claude Code 中验证
!echo $ANTHROPIC_AUTH_TOKEN

Protect credentials

Keep API keys out of Git and public conversations, and rotate them as needed. The tutorial illustrates environment variables:

bash

# 更安全的方式：使用环境变量
export ANTHROPIC_AUTH_TOKEN="sk-ant-xxxxx"

Exclude sensitive local configuration from version control:

# 不要提交敏感配置
settings.json
.env
.env.local

Limit local file permissions on Linux/macOS:

bash

# Linux/macOS 设置文件权限
chmod 600 ~/.claude.json

Advanced usage

The following sections cover MCP integrations and specialized Skills for developers already comfortable with the basic workflow.

Chapter 1: MCP integrations

1.1 Configure and manage MCP servers

What is MCP?

Model Context Protocol connects the assistant to external tools and data sources through MCP servers.

Configuration location

The source tutorial uses the ModelScope MCP directory to find services.

bash

# Windows
%USERPROFILE%\.claude.json

Basic configuration

json

{
  "mcpServers": {
    "bing-search": {
      "command": "npx",
      "args": ["-y", "bing-cn-mcp"]
    }
  }
}

1.2 Example MCP services

Bing search in Chinese

json

{
  "mcpServers": {
    "bing-search": {
      "command": "npx",
      "args": ["-y", "bing-cn-mcp"]
    }
  }
}

Example request:

使用必应搜索最新的 AI 新闻

Search results can include titles, links, and summaries. Further page retrieval, pagination, and filters depend on the functions exposed by the selected server.

Several servers

json

{
  "mcpServers": {
    "bing-search": {
      "command": "npx",
      "args": ["-y", "bing-cn-mcp"]
    },
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelscope/mcp-filesystem"]
    },
    "postgresql": {
      "command": "npx",
      "args": ["-y", "@modelscope/mcp-postgresql"]
    }
  }
}

1.3 Inspect available tools

Ask in the conversation:

你有哪些 MCP 工具？

Check the configured servers and the functions actually available to the current session.

1.4 Practical uses

Task	Example MCP server	Use
Current information	`bing-search`	Search news, prices, or weather
Database work	`postgresql`	Query or update according to the server's available tools and database permissions
File access	`filesystem`	Access explicitly configured directories
API integration	`custom-api`	Connect internal business APIs

1.5 Configuration practices

Pass credentials through supported environment configuration and limit each server's access. Configure only the services needed, then observe invocation latency and success rates. Keep server versions maintained and retain appropriate logs for important calls.

Chapter 2: Skills

2.1 Overview

Skills provide instructions and supporting tools for specialized file types or tasks. The source tutorial demonstrates three skill packages; these are examples, not capabilities automatically bundled with every installation.

Three example skills

Skill	Tasks	Example dependencies
`ai-tutor`	Technical explanations and video transcription	Python, FFmpeg
`pdf`	PDF creation, editing, forms, and extraction	LibreOffice, ReportLab
`xlsx`	Spreadsheets, analysis, and charts	LibreOffice, openpyxl, pandas

2.2 Installation and dependencies

Step 1: download the skills

bash

# 克隆 Skills 仓库
git clone https://github.com/anthropics/skills.git

# 或访问在线浏览
https://github.com/anthropics/skills/tree/main/skills

Step 2: copy them into the Claude directory

bash

# Windows
复制到 %USERPROFILE%\.claude\skills\

# macOS/Linux
复制到 ~/.claude/skills/

Step 3: install dependencies

Python:

bash

# 安装最新 Python（版本 3.10+）
# Windows: https://www.python.org/downloads/
# macOS: brew install python@3.11
# Linux: apt-get install python3.11

# 验证安装
python --version

LibreOffice:

bash

# Windows: 下载完整包
https://zh-cn.libreoffice.org/download/libreoffice/

# macOS
brew install libreoffice

# Linux
apt-get install libreoffice

Python packages:

bash

pip install reportlab -q
pip install openpyxl -q
pip install pandas -q
pip install pillow -q
pip install python-pptx -q

Verify installation:

bash

# 在 Claude Code 中查询可用 Skills
你有哪些 Skills？

2.3 ai-tutor

Examples

Explain a technical concept:

ultrathink: 请用 ai-tutor 给我讲解什么是微服务架构，
包括核心概念、优缺点和应用场景

Transcribe and analyze a video:

帮我转录这个视频文件，并总结核心要点
@/path/to/video.mp4

Usage tips

Supply relevant reference material and request the depth you need. The source examples use ultrathink: where supported. Divide long videos into sections for easier processing and review.

2.4 PDF

Main tasks

Create a PDF:

帮我创建一个项目总结报告 PDF，
包含：项目概述、技术方案、进度统计、团队成员信息

Fill a form:

帮我填充这个 PDF 表单：
@表单模板.pdf

数据内容：
- 姓名：张三
- 日期：2026-03-04
- 签名：已确认

Extract text and tables:

请提取这个 PDF 中的所有表格数据，
转换为 CSV 格式
@/path/to/document.pdf

Merge and split documents:

合并这些 PDF 文件：
@file1.pdf
@file2.pdf
@file3.pdf

输出为：merged_document.pdf

Advanced examples

Batch processing:

bash

# 处理目录下所有 PDF
帮我批量提取 ./pdf_folder 中所有 PDF 的文本内容

Conditional conversion:

将这个 PDF 转换为图片，
仅保留第 1-5 页
@source.pdf

2.5 XLSX

Main tasks

Create structured data:

创建一个产品销售统计表，
包含：产品名称、销售数量、销售额、利润率、同比增长
示例数据：5条产品记录

Analyze data and create charts:

分析这个 Excel 文件，
生成销售趋势图表和关键指标总结
@sales_data.xlsx

Add formulas and calculations:

创建财务预算表，
包含收入、成本、利润等项目，
自动计算小计和总计

Build pivot tables and statistics:

基于这个数据文件，
按地区和产品类别生成销售汇总表
@detailed_sales.xlsx

Spreadsheet practices

Use clear column headings, appropriate number formats or conditional formatting, and data-validation rules to reduce input errors.

AI-Assisted Programming ​

Getting started ​

Prerequisites ​

Install Claude Code ​

1. VS Code extension ​

2. JetBrains integration ​

1. Initialize project context ​

2. Switch permission modes ​

When to use Plan Mode ​

Choose a mode at startup ​

3. Reference files and directories ​

4. Export a conversation ​

5. Request deeper reasoning ​

6. Paste screenshots ​

7. Correct direction early ​

8. Navigate previous prompts ​

9. Run a shell command ​

10. Search prompt history ​

11. Skip permission checks with care ​

Appendix: configuration examples ​

Configuration locations ​

.claude.json ​

Complete example ​

Fields in the example ​

settings.json ​

Location ​

Complete example ​

Environment fields ​

Configuration steps ​

Step 1: create .claude.json ​

Step 2: optionally create settings.json ​

Step 3: verify configuration ​

Protect credentials ​

Advanced usage ​

Chapter 1: MCP integrations ​

1.1 Configure and manage MCP servers ​

What is MCP? ​

Configuration location ​

Basic configuration ​

1.2 Example MCP services ​

Bing search in Chinese ​

Several servers ​

1.3 Inspect available tools ​

1.4 Practical uses ​

1.5 Configuration practices ​

Chapter 2: Skills ​

2.1 Overview ​

Three example skills ​

2.2 Installation and dependencies ​

Step 1: download the skills ​

Step 2: copy them into the Claude directory ​

Step 3: install dependencies ​

2.3 ai-tutor ​

Examples ​

Usage tips ​

2.4 PDF ​

Main tasks ​

Advanced examples ​

2.5 XLSX ​

Main tasks ​

Spreadsheet practices ​

AI-Assisted Programming

Getting started

Prerequisites

Install Claude Code

1. VS Code extension

2. JetBrains integration

1. Initialize project context

2. Switch permission modes

When to use Plan Mode

Choose a mode at startup

3. Reference files and directories

4. Export a conversation

5. Request deeper reasoning

6. Paste screenshots

7. Correct direction early

8. Navigate previous prompts

9. Run a shell command

10. Search prompt history

11. Skip permission checks with care

Appendix: configuration examples

Configuration locations

.claude.json

Complete example

Fields in the example

settings.json

Location

Complete example

Environment fields

Configuration steps

Step 1: create .claude.json

Step 2: optionally create settings.json

Step 3: verify configuration

Protect credentials

Advanced usage

Chapter 1: MCP integrations

1.1 Configure and manage MCP servers

What is MCP?

Configuration location

Basic configuration

1.2 Example MCP services

Bing search in Chinese

Several servers

1.3 Inspect available tools

1.4 Practical uses

1.5 Configuration practices

Chapter 2: Skills

2.1 Overview

Three example skills

2.2 Installation and dependencies

Step 1: download the skills

Step 2: copy them into the Claude directory

Step 3: install dependencies

2.3 ai-tutor

Examples

Usage tips

2.4 PDF

Main tasks

Advanced examples

2.5 XLSX

Main tasks

Spreadsheet practices