Guardrails Design

Design guardrails for this Custom GPT:

GPT PURPOSE:
[What it does]

POTENTIAL RISKS:
[What could go wrong]

GUARDRAILS NEEDED:

CONTENT BOUNDARIES:
- Topics to avoid entirely:
- Topics to handle carefully:
- When to recommend professional help:

BEHAVIORAL LIMITS:
- What the GPT should refuse to do:
- How to decline gracefully:
- What to do instead:

ACCURACY PROTECTION:
- When to express uncertainty:
- When to recommend verification:
- What not to guess about:

USER PROTECTION:
- Privacy considerations:
- Safety considerations:
- Emotional considerations:

Write guardrail instructions for the system prompt that are firm but not so restrictive they break the GPT's usefulness.