Tips & Best Practices

Master OpenClaw with proven strategies for prompt engineering, performance optimization, security, and production deployment.

Prompt Engineering Tips

Write effective system prompts:

# Good: Specific, actionable instructions
system_prompt: |
  You are a Python code reviewer.
  - Focus on PEP 8 compliance
  - Identify security vulnerabilities
  - Suggest performance improvements
  - Provide code examples for fixes

# Bad: Vague, general instructions
system_prompt: "You are helpful assistant."

Use context efficiently:

• Keep recent messages relevant (summarize old context)
• Include file paths and line numbers in questions
• Break complex tasks into smaller steps

Performance Optimization

Reduce Latency

performance:
  response_streaming: true  # Stream responses
  parallel_tool_calls: true  # Execute tools concurrently
  cache_embeddings: true
  cache_ttl: 3600

Optimize Token Usage

• Use cheaper models for simple tasks (gpt-3.5-turbo)
• Implement context pruning for long conversations
• Cache frequently-used responses

optimization:
  context_management:
    max_tokens: 8000
    summarize_after: 6000
  response_caching:
    enabled: true
    ttl: 1800

Security Best Practices

Protect Sensitive Data

security:
  secrets_management:
    provider: "vault"  # or "aws_secrets_manager"
    auto_rotate: true

  access_control:
    require_authentication: true
    allowed_users:
      - "user_id_1"
      - "user_id_2"

  audit_logging:
    enabled: true
    log_level: "info"
    sensitive_fields_mask: true

Sandbox Restrictions

security:
  code_execution:
    network_access: false
    file_system_access: "read-only"
    timeout: 30000

  tool_restrictions:
    blacklist_commands:
      - "rm -rf"
      - "dd if="
      - ":(){ :|:& };:"

Production Deployment Checklist

• Set up monitoring and alerts (Prometheus, Grafana)
• Configure log aggregation (ELK stack, Loki)
• Enable backup automation (database, configs)
• Set resource limits (CPU, memory, disk)
• Configure rate limiting and quotas
• Set up SSL/TLS certificates
• Implement health checks and readiness probes
• Document incident response procedures
• Test disaster recovery plan
• Review and update security policies

# production-config.yaml
environment: "production"
log_level: "warn"
monitoring:
  prometheus:
    enabled: true
    port: 9090
backup:
  schedule: "0 2 * * *"
  retention_days: 30