Conversational deployment of OpenClaw AI assistant to VPS server using Claude Code
中文 | English
- What's Included
- Quick Start
- System Requirements
- Features After Deployment
- Usage Examples
- Recommended Skills and Extensions
- Documentation Structure
- Common Commands
- Troubleshooting
- Security Recommendations
- FAQ
- Resources
- Contributing
- Author
- License
One-Click Installation Script (install-openclaw.sh)
- Auto-install all dependencies
- Interactive/non-interactive configuration
- Auto-start service
- Config backup and restore
# Download script
wget https://raw.githubusercontent.com/taielab/openclaw-autopilot/main/install-openclaw.sh
# Make executable
chmod +x install-openclaw.sh
# Run installation
./install-openclaw.sh-
Prepare the following information:
- VPS server IP and SSH credentials
- Telegram Bot Token
- AI API key
-
Send to Claude Code:
I want to deploy OpenClaw AI assistant on VPS: - Server: <IP> - User: root - Password: <password> - Telegram Bot Token: <token> - API: Yunwu AI, Key: <api-key> Please help me complete the automated deployment. -
Claude Code will automatically complete all deployment steps
- OS: Ubuntu 22.04 LTS (recommended)
- Specs: Minimum 2 cores, 2GB RAM
- Network: International network access
- Permissions: root or sudo access
- 15 top-tier AI models (GPT-5.2, Claude Opus 4.5, DeepSeek V3.2, etc.)
- Telegram Bot interface
- Automatic failover
- Browser automation
- Multiple skills and tools
- AI/ML: Whisper, yt-dlp, Pandas, NumPy
- Development: GitHub CLI, Git, Node.js, Python3
- Media: FFmpeg, ImageMagick
- System: jq, ripgrep, fd, bat, curl, wget
- 🐙 GitHub operations
- 🎞️ Video processing
- 🌤️ Weather queries
- 🧵 tmux control
- 📦 Skill creation
Find your Bot on Telegram and send a message:
Hello! Help me analyze this project: https://github.com/openclaw/openclaw
AI will automatically use GitHub CLI tools to fetch repository information and analyze it.
/model
Bot will display all 15 available models for you to choose from.
Download this video and extract key frames: https://youtube.com/watch?v=xxxxx
AI will use yt-dlp and FFmpeg to process automatically.
Calculate the first 20 Fibonacci numbers using Python
AI will write and execute Python code, returning the results.
Get the main content from https://example.com
AI will use browser automation tools to scrape web content.
OpenClaw supports rich skill extensions. Here are community-recommended quality skills:
-
Feishu Integration: clawdbot-feishu
https://github.com/m1heng/clawdbot-feishu -
WeChat Integration: OpenClaw-Wechat
https://github.com/dingxiang-me/OpenClaw-Wechat -
DingTalk Integration: dingtalk-moltbot-connector
https://github.com/DingTalk-Real-AI/dingtalk-moltbot-connector -
China Optimization: moltbot-china
https://github.com/BytePioneer-AI/moltbot-china
- Awesome Skills: awesome-openclaw-skills
https://github.com/VoltAgent/awesome-openclaw-skills
💡 Tip: Installing skills can greatly expand your AI assistant's capabilities, supporting more scenarios and workflows.
.
├── README.md # Chinese version
├── README_EN.md # English version (this file)
└── install-openclaw.sh # One-click installation script
# Check service status
systemctl --user status openclaw-gateway
# Start service
systemctl --user start openclaw-gateway
# Stop service
systemctl --user stop openclaw-gateway
# Restart service
systemctl --user restart openclaw-gateway
# View logs
journalctl --user -u openclaw-gateway -f# Health check
openclaw doctor
# List skills
openclaw skills list
# List models
openclaw models list
# Test configuration
openclaw gateway --port 18789# View detailed errors
openclaw gateway --port 18789
# Check configuration
openclaw doctor
# View logs
journalctl --user -u openclaw-gateway -n 50# Test Bot Token
curl https://api.telegram.org/bot<YOUR_TOKEN>/getMe
# Check network
curl https://api.telegram.org
# View Gateway logs
journalctl --user -u openclaw-gateway -f# Test API connection
curl -X POST https://yunwu.ai/v1/chat/completions \
-H "Authorization: Bearer <YOUR_KEY>" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-5.2","messages":[{"role":"user","content":"test"}]}'-
Restrict Access
- Modify
allowFromto allow only specific users - Use
dmPolicy: "pairing"to enable pairing mode
- Modify
-
Use Firewall
ufw allow from 127.0.0.1 to any port 18789
-
Regular Updates
npm update -g openclaw apt-get update && apt-get upgrade -
Backup Configuration
tar -czf openclaw-backup-$(date +%Y%m%d).tar.gz ~/.openclaw
A: 15 top-tier models:
- OpenAI: GPT-5.2, GPT-5.2 Codex
- Anthropic: Claude Opus 4.5, Claude Sonnet 4.5 (with Thinking versions)
- DeepSeek: DeepSeek V3.2 (with Thinking version)
- Google: Gemini 3 Pro, Gemini 3 Pro Image
- Zhipu: GLM-4.7 (with Thinking version)
- MiniMax: MiniMax M2.1
- Alibaba: Qwen3 Max
- xAI: Grok 4.1
A: Main costs include:
- VPS Server: ~$5-10/month (2 cores, 2GB RAM)
- API Calls: Pay-as-you-go, using Yunwu AI proxy can reduce costs
- Telegram Bot: Free
A: Yes, but you need to:
- Keep your computer running 24/7
- Configure port forwarding (if external access needed)
- VPS is recommended for better stability
A: Edit ~/.openclaw/openclaw.json, add new provider configuration in models.providers, and add corresponding authentication in auth-profiles.json. See complete tutorial for details.
A: OpenClaw supports multiple channels:
- Telegram (configured in this tutorial)
- Discord
- Slack
- Signal
- iMessage
See official documentation for configuring other channels.
A: Run the following commands:
npm update -g openclaw
systemctl --user restart openclaw-gatewayA:
2. Run openclaw doctor for health check
3. View logs journalctl --user -u openclaw-gateway -f
4. Submit an Issue on GitHub
5. Join Discord community for help
- OpenClaw Official Docs: https://docs.openclaw.ai
- OpenClaw GitHub: https://github.com/openclaw/openclaw
- This Project GitHub: https://github.com/taielab/openclaw-autopilot
- Discord Community: https://discord.gg/clawd
- Yunwu AI: https://yunwu.ai/register?aff=PBpy (Recommended API Provider)
Issues and Pull Requests are welcome!
- Fork this repository
- Create a feature branch (
git checkout -b feature/AmazingFeature) - Commit your changes (
git commit -m 'Add some AmazingFeature') - Push to the branch (
git push origin feature/AmazingFeature) - Open a Pull Request
- Ensure code passes tests before submitting
- Follow existing code style
- Update relevant documentation
- Describe changes in detail in PR
AI Security Workshop
Follow us for more AI tech content:
- WeChat Official Account: AI安全工坊
- X (Twitter): https://x.com/JackW_AGI
- GitHub: https://github.com/taielab
MIT License
Enjoy!
If you have any questions, please check the complete tutorial or submit an issue on GitHub.
New
- Gateway bind customization: loopback / 0.0.0.0 / custom IP
- Gateway port customization
- Environment variables:
GATEWAY_BIND,GATEWAY_PORT
🚀 Script Features
- Non-interactive mode (
-n) + environment variables - Update mode (
-u) preserves config - Config backup + DELETE confirmation
- Telegram allowlist configuration
- CLI auto-completion
⚙️ Configuration (93% coverage)
- Agent: timeout(300s), mediaMax(20MB), thinking(low), compaction
- Session: reset(daily/3am), dmScope(per-channel-peer), typingMode
- Telegram: streamMode(partial), groupPolicy(allowlist)
- System: env, cron, web, canvasHost, logging, diagnostics
🔐 Security
- Allowlist mode auto-sets dmPolicy=pairing
- Config file permissions chmod 600
- Session isolation per-channel-peer
- Initial release