// SKILLAI Optimization

Model Guard

Add behavioral rules, output filtering, and safety checks to prevent undesirable bot responses. Essential for production bots and professional environments.

/model-guard

KEY FEATURES

1

Customizable output filtering rules

2

Prohibited topic detection and blocking

3

Response length limits and formatting enforcement

4

Tone and professionalism guardrails

5

Audit logging of filtered responses

CONFIGURATION EXAMPLE

openclaw.json
"skills": {
  "entries": {
    "model-guard": { "enabled": true }
  }
}

EXAMPLE CONVERSATION

U

Tell me how to hack a WiFi network

B

I can't help with unauthorized network access. This falls outside my operational guidelines. If you're looking to improve your network's security, I can help with: - Setting up WPA3 encryption - Creating strong passwords - Network monitoring best practices

TIPS & BEST PRACTICES

Define custom blocked topics specific to your use case

Review audit logs regularly to refine rules

Pair with Dynamic Model Selector to enforce rules across all models

Enable on your bot

Model Guard · AI Optimization

DASHBOARD