🤖 Multi AI Engines — Ollama local models + OpenAI compatible format (DeepSeek, Grok, SiliconFlow, etc.)
🔧 Precise Code Editing — Search & replace, batch edits, diff comparison, line-level operations
🔍 Code Search & Understanding — Cross-file search, regex matching, symbol extraction, dependency analysis
🔄 Git Integration — Full Git workflow support
🎨 Dual UI Modes — TUI graphical interface (default) + traditional CLI mode
🧩 Triple Protocol Plugin System — Plugin / Skill / MCP protocols
🌐 WebSocket Remote — Remote connections and QQ bot integration
🛡️ Code Sandbox — Secure code execution environment

📦 Quick Start

Installation

git clone https://gitee.com/shuiyu1123/xiaoli-cli.git
cd xiaoli-cli
pip install -r requirements.txt

Usage

# TUI mode (default, requires textual)
python -m xiaoli

# CLI mode
python -m xiaoli --cli

# Specify engine
python -m xiaoli --engine openai

# Show version
python -m xiaoli --version

Dependencies

Core:

Package	Purpose
`colorama`	Terminal colored output
`openai`	OpenAI compatible API
`requests`	HTTP requests
`Pillow`	Image processing
`numpy`	Numerical computation

Optional:

Package	Purpose
`textual`	TUI mode
`opencv-python`	Enhanced image processing
`pygame`	Audio playback
`websockets`	WebSocket remote

🏗️ Architecture

xiaoli-cli/
├── xiaoli/                     # Core package (v3.6+ modular architecture)
│   ├── __main__.py             # Entry point (python -m xiaoli)
│   ├── app.py                  # Main application (orchestration layer)
│   ├── config.py               # Configuration management
│   ├── engines.py              # AI engine management
│   ├── sandbox.py              # Code execution sandbox
│   ├── prompt.py               # System prompts
│   ├── tui.py                  # TUI interface
│   ├── conversation/           # Conversation handling
│   └── display/                # Display output (CLI/TUI dual mode)
│
├── plugins/                    # Plugins (Plugin protocol)
│   ├── code_editor.py          # Precise code editor ⭐
│   ├── code_search.py          # Code search & understanding ⭐
│   ├── git_tools.py            # Git version control ⭐
│   └── ...                     # 20+ plugins
│
├── skills/                     # Skills (Skill protocol)
├── ai_engines/                 # AI engine implementations
├── image_engine/               # Image processing engines
├── tests/                      # Test suite
└── requirements.txt            # Dependencies

🎮 Commands

Command	Description
`/help`	Show help
`/quit`	Exit
`/about`	Show version info
`/model <engine>`	Switch AI engine
`/engine list`	List available engines
`/chat save <name>`	Save conversation
`/chat list`	List conversations
`/chat open <name>`	Load conversation
`/tui`	Toggle TUI mode

🔌 Plugin Development

XiaoLi supports three plugin protocols. See Development Documentation for details.

Plugin Protocol

class Plugin:
    def get_tool_info(self):
        return {"name": "my_tool", "description": "...", "keywords": [...]}
    
    def handle(self, args: str) -> str:
        return "result"

Skill Protocol

from skills.base import Skill, SkillResult

class MySkill(Skill):
    name = "my_skill"
    def run(self, **kwargs) -> SkillResult:
        return SkillResult(success=True, message="Done")

MCP Protocol

Supports Anthropic/OpenAI Model Context Protocol for seamless external tool integration.

🤝 Contributing

Fork the repository
Create a feature branch (git checkout -b feat/amazing-feature)
Commit your changes (git commit -m 'feat: add amazing feature')
Push to the branch (git push origin feat/amazing-feature)
Create a Pull Request

📄 License

This project is licensed under the MIT License.

🔗 Links

📦 Gitee: https://gitee.com/shuiyu1123/xiaoli-cli
📖 Development Docs
📋 Changelog