Use this page as a reference sheet. Once you begin deploying, testing models, and debugging channels, this page is usually faster than re-reading full chapters.
A. FAQ
Q1. Is OpenClaw free?
The software is MIT-licensed and free, but using it usually means paying for either infrastructure, model usage, or both.
Q2. Do I need advanced technical skills?
Not necessarily for the basics. But once you start connecting many platforms, building Skills, and tightening security, technical competence becomes more important.
Q3. What is the biggest difference from ChatGPT?
ChatGPT is mainly conversational. OpenClaw is more about persistent task execution inside your own environment.
Q4. Is it safe?
It can be operated safely, but not automatically. Authentication, careful Skill installation, and budget controls matter a lot.
Q5. How much does it cost per month?
That depends heavily on model choice. Local models can be nearly free. A careful DeepSeek or GLM setup can stay low-cost. Premium Claude-heavy usage can climb fast without controls.
Q6. Can I use domestic Chinese models?
Yes. DeepSeek + GLM is one of the most practical combinations for many Chinese users.
Q7. Why is Claude so often used as the main model?
Because Claude remains one of the strongest options for agentic reasoning, multi-step tool use, and stability in harder workflows.
Q8. Is the project still worth using after the founder joined OpenAI?
Yes, as long as you are comfortable with an open-source system that still moves quickly and changes often.
Q9. Are Skills in ClawHub safe?
Do not assume so. Review source and install conservatively.
Q10. Can it connect directly to personal WeChat?
There are routes, but they are higher-risk and less stable than mature platforms such as QQ, Feishu, DingTalk, or WeCom.
Q11. Can OpenClaw and Claude Code be used together?
Yes. Many advanced workflows benefit from exactly that split.
Q12. Are local models worth it?
Absolutely for privacy, testing, and cost-sensitive experimentation. Just do not expect them to always match the strongest cloud models on complex agent tasks.
B. Command Cheat Sheet
Install and update
| Command | Meaning |
|---|---|
npm install -g openclaw@latest | Install globally |
openclaw onboard --install-daemon | Initial setup plus daemon install |
openclaw update --channel stable | Update stable |
openclaw update --channel beta | Update beta |
openclaw doctor | Diagnostics |
openclaw --version | Show version |
Daily usage
| Command | Meaning |
|---|---|
openclaw gateway --port 18789 --verbose | Start Gateway |
openclaw gateway restart | Restart after config changes |
openclaw agent --message "xxx" | Send a message directly |
openclaw devices pair | Pair a new device |
openclaw models list | List models |
openclaw models status --probe | Probe connectivity |
openclaw config set agents.defaults.model.primary provider/model | Set primary model |
Plugins and channels
| Command | Meaning |
|---|---|
openclaw plugins install <name> | Install a plugin |
openclaw plugins enable <name> | Enable a plugin |
openclaw plugins list | List plugins |
openclaw plugins install @openclaw-china/channels | Install China channel pack |
openclaw china setup | Configure China-focused messaging platforms |
Model auth
| Command | Meaning |
|---|---|
openclaw onboard --auth-choice zai-api-key | Configure Zhipu |
openclaw onboard --auth-choice apiKey --token-provider openrouter --token "$KEY" | Configure OpenRouter |
openclaw models auth login --provider qwen-portal --set-default | Configure Qwen portal login |
In-chat commands
| Command | Meaning |
|---|---|
/status | Show session state |
/new | Start a new session |
/think <level> | Adjust reasoning depth |
| `/usage off | tokens |
| `/activation mention | always` |
Docker
| Command | Meaning |
|---|---|
docker-compose up -d | Start in background |
docker-compose logs -f | Follow logs |
docker-compose pull | Pull fresh images |
C. Resource Links
Official resources
| Name | Link |
|---|---|
| GitHub repo | github.com/openclaw/openclaw |
| Docs | docs.openclaw.ai |
| Website | openclaw.ai |
| ClawHub | clawhub.ai |
| Releases | github.com/openclaw/openclaw/releases |
| Discussions | github.com/openclaw/openclaw/discussions |
Community resources
| Name | Why it matters |
|---|---|
awesome-openclaw-skills | Curated Skill discovery |
awesome-openclaw-usecases | Real-world examples |
openclaw-claude-code-skill | Claude Code bridge |
SecureClaw | Skill security scanning |
China-focused resources
| Name | Why it matters |
|---|---|
openclaw-china | Mainland messaging integrations |
| Chinese docs mirrors | Local-language guidance |
| Alibaba Cloud deployment docs | Template-based deployment |
| Bilibili tutorials | Beginner-friendly walkthroughs |
| Zhihu and blog guides | Troubleshooting and deployment notes |
Model consoles
| Provider | Console |
|---|---|
| Anthropic | console.anthropic.com |
| OpenAI | platform.openai.com |
| Google AI Studio | aistudio.google.com |
| DeepSeek | platform.deepseek.com |
| Zhipu | bigmodel.cn |
| Qwen | dashscope.aliyun.com |
| Moonshot | platform.moonshot.cn |
| SiliconFlow | siliconflow.cn |
| OpenRouter | openrouter.ai |
| Volcengine | console.volcengine.com |