Your First OpenClaw Config File
Skip the Learning Curve
Our Config Wizard generates production-ready configs in 30 seconds. No YAML knowledge required.
Generate Config NowWhat is clawhub.json?
The clawhub.json file is your agent's brain. It tells OpenClaw:
- Which LLM to use (OpenAI, DeepSeek, Claude)
- Your API credentials
- Agent behavior and capabilities
- Plugin configurations
Minimal Example
Here's the simplest possible config:
{
"llm": {
"provider": "openai",
"apiKey": "sk-...",
"model": "gpt-4o"
}
}Recommended Config (Production-Ready)
For real-world use, add these fields:
{
"llm": {
"provider": "openai",
"apiKey": "sk-...",
"model": "gpt-4o",
"temperature": 0.7,
"maxTokens": 4000
},
"agent": {
"name": "MyAssistant",
"host": "127.0.0.1",
"port": 3000,
"logLevel": "info"
},
"skills": [
"browser-use",
"memory-core"
]
}Using DeepSeek (10x Cheaper)
DeepSeek V3.2 offers GPT-4-level performance at 1/10th the cost:
{
"llm": {
"provider": "deepseek",
"apiKey": "sk-...",
"model": "deepseek-chat",
"baseURL": "https://api.deepseek.com/v1"
}
}💡 Pro Tip: Use our Cost Estimator to compare pricing before committing.
Common Mistakes to Avoid
❌ Trailing Commas
JSON doesn't support trailing commas. Our Config Wizard handles this automatically.
❌ Windows Path Escaping
Use C:\\\\Users not C:\\Users. The wizard auto-fixes this.
❌ Hardcoded API Keys
Never commit API keys to Git. Use environment variables instead.
Best Practices
Use 127.0.0.1 instead of localhost
Avoids IPv6 issues in Node.js v18+
Set maxTokens to prevent runaway costs
Recommended: 4000 for most tasks
Enable logging in development
Set logLevel to "debug" for troubleshooting
Validate before deploying
Use Config Wizard to catch syntax errors
Next Steps
Last Updated: February 6, 2026 | Time to Create: 30 seconds (with Config Wizard) or 10 minutes (manual)