xiaoluolyg--god/jiuwenclaw/README.md

JiuwenClaw

Your On-Call AI Butler — Bringing Intelligence to Your Fingertips

🌟 Overview

JiuwenClaw is an intelligent AI Agent built in Python. True to its name — "Claw" symbolizes precise reach and connection — it extends the power of large language models directly to your fingertips through the communication apps you already use every day.

✨ Key Features

• Ecosystem Compatible: Full support for Huawei Cloud MaaS and other mainstream model platforms
• Seamless Integration: Native integration with the Xiaoyi Open Platform, enabling Huawei phone users to invoke JiuwenClaw directly through the Xiaoyi assistant
• Flexible Deployment: Self-hosted deployment with full data sovereignty
• Multi-Platform Access: Interact via web interface, messaging apps, and more

🎯 Design Philosophy

Understands You. Evolves With You.

🤝 Your Personal Task Butler

Whether dealing with task additions, mid-flow interruptions, or shifting requirements, JiuwenClaw understands your intent precisely — intelligently scheduling and executing tasks in an orderly, stress-free manner.

🔄 Autonomous Evolution

When you express dissatisfaction or an error occurs, JiuwenClaw automatically refines the relevant skills based on your feedback — continuously improving, always working in your best interest.

⚡ Your always-on, data-sovereign personal AI assistant ⚡

⚠️ Version Upgrade Notice

If you're upgrading from an older version, check the changelog for any breaking changes. You must reinitialize JiuwenClaw if a breaking change is indicated. The service will fail to start without reinitialization.

Backup Before Upgrading

Data Type	Source Path	Description
Memory Data	.jiuwenclaw/workspace/agent/memory	All your conversation memories
Custom Skills	.jiuwenclaw/workspace/agent/skills	Your custom agent skills
Configuration	.jiuwenclaw/config	Your app settings

Migration Steps

After upgrading and running jiuwenclaw-init, manually migrate your data:

1. Copy Memory:

   cp -r .jiuwenclaw/workspace/agent/memory .jiuwenclaw/agent/memory
   

2. Copy Skills:

   cp -r .jiuwenclaw/workspace/agent/skills .jiuwenclaw/agent/skills
   

🚀 Getting Started

📦 Installation

# Install JiuwenClaw
pip install jiuwenclaw

# Initialize JiuwenClaw (first-time setup or after upgrading)
# ⚠️ Remember to backup your data before running this command
jiuwenclaw-init

# Start JiuwenClaw
jiuwenclaw-start

# Install JiuwenClaw-tui
pip install jiuwenclaw-tui

# Start JiuwenClaw-tui
jiuwenclaw-tui

💬 How to Use

1️⃣ Conversation Mode

Method	Description
Web Frontend	After starting the service, visit http://localhost:5173 to chat directly in your browser
Xiaoyi Channel	Huawei phone users can invoke Xiaoyi to talk with JiuwenClaw directly
Lark Channel	Once configured, chat with JiuwenClaw seamlessly inside Lark

2️⃣ Scheduled Tasks

Set up heartbeat tasks with your to-do items, and JiuwenClaw will wake up on schedule to execute them automatically — making your time management smarter and more effortless.

📚 Documentation

Document	Description
📖 Install guide	Full installation paths (pip, source, conda, Docker)
📖 Quick Start	Get up and running with JiuwenClaw in 5 minutes
📖 Quick Start (TUI)	Get up and running with JiuwenClaw-tui in 5 minutes
⚙️ Configuration & Workspace	Environment setup and workspace management
📁 Workspace Structure	workspace directory layout, presets, and dynamic content
🔄 Modes	PLAN / AGENT / CODE / TEAM mode switching and configuration
🛠️ Skill System	Guide to developing custom skills
🔄 Skill Self-Evolution	Mechanism for automatic skill evolution
📱 Channel Configuration	Integrating Feishu, Xiaoyi, and other channels
💬 Discord	Discord channel setup and usage
💬 WhatsApp	WhatsApp channel setup and usage
⌨️ CLI Commands	Command-line tool usage guide
⏰ Scheduled Tasks	Scheduled task management
💓 Heartbeat	Heartbeat mechanism and configuration
🧠 Memory	Intelligent memory and learning capabilities
💡 Task Memory	Task-level experience retrieval and consolidation
📦 Context Compression	Context compression and unloading
💻 Coding Memory	Code-mode-specific memory system
📋 Task Planning	Chat behavior and task flow
🌐 Browser Automation	Web browsing and automation features
🔌 MCP Configuration	MCP server integration and configuration
🔒 Tool Permissions & Security	Permission model and security configuration
📝 Slash Commands	Slash command reference
🏗️ Slash Command Architecture	Slash command internals and extension guide
📨 E2A Protocol	Gateway ↔ Agent request envelope specification
🤝 A2A Integration	A2A protocol integration guide
🔌 ACP Client Config	ACP client plugin configuration
👥 Distributed Team	Multi-process distributed team mode
🔀 Multi-Instance Operation	Running multiple independent instances on one machine
📦 Packaging Desktop App	Build standalone desktop executables
🚀 Development Practices	Development practices and experience sharing

🤝 Contributing

We warmly welcome community contributions — whether it's filing bug reports, suggesting new features, or improving documentation, every bit of support means the world to us.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

---

Making intelligence accessible. Making life simpler.
_{✨ JiuwenClaw — Your Personal AI Assistant ✨}