U-Claw provides an offline, portable USB installer for the OpenClaw AI assistant, specifically tailored for users in China. It addresses the complexities of setting up AI tools by offering a one-click, no-VPN solution, enabling users to run a fully functional AI environment directly from a USB drive with a familiar "Yulin Mufeng" style interface. This project benefits users seeking a convenient and accessible AI deployment, particularly those facing network restrictions.

How It Works

The core of U-Claw is a repository containing the necessary scripts and configurations to create a portable OpenClaw environment. A setup.sh script handles the download and installation of essential large dependencies like Node.js and OpenClaw itself, utilizing domestic mirrors to bypass potential network issues. Once dependencies are met, the portable/ directory holds the complete, self-contained application ready to be copied onto a USB drive, allowing for direct execution via platform-specific batch or command files without altering the host system.

Quick Start & Requirements

• Primary Install/Run:
  1. git clone https://github.com/dongsheng123132/u-claw.git
  2. cd u-claw/portable
  3. bash setup.sh (Installs Node.js and OpenClaw dependencies using domestic mirrors, approx. 1 minute)
  4. Copy the portable/ directory to a USB drive.
  5. Run from USB: Mac-Start.command (macOS) or Windows-Start.bat (Windows).
• Prerequisites: Node.js, OpenClaw (downloaded by setup.sh). Specific runtime environments include node-mac-arm64 and node-win-x64.
• Resource Footprint: USB drive recommended: 4GB+ (full version approx. 2.3GB).
• Links: Official Website: u-claw.org

Highlighted Details

• Offline Functionality: Designed for offline use, with all necessary components packaged for direct execution from a USB drive.
• Domestic AI Model Support: Prioritizes and integrates Chinese AI models (DeepSeek, Kimi K2.5, Qwen, GLM, MiniMax, Doubao) that do not require VPN access.
• Multi-Platform Chat Integration: Includes built-in support for QQ, Feishu, Telegram, WhatsApp, and Discord, with a community plugin for WeChat.
• Desktop Application: An alternative Electron-based desktop application version is also available.

Maintenance & Community

The project is maintained by dongsheng (@dongsheng123132 on GitHub). Contact is available via WeChat (hecare888). Areas for future development and community contributions include enhancing Windows portable testing, adding Mac Intel and Linux support (AppImage), implementing desktop auto-updates, refining the SkillHub feature, and adding multi-language support (English UI).

Licensing & Compatibility

The project is released under the MIT License, permitting free copying and distribution. This license generally allows for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

Current platform support primarily targets macOS (Apple Silicon) and Windows (64-bit); Mac Intel and Linux support are listed as under development. International AI models (Claude, GPT, Gemini) necessitate a VPN or proxy connection. WeChat integration relies on a community-developed plugin. macOS users may encounter "unverified developer" warnings when running scripts, requiring manual override (e.g., right-click -> Open).