Prompting

Your system prompt defines the agent’s role, tone, and limits. Keep it short, explicit, and easy to maintain.

Suggested structure

• Role: What the agent is responsible for.
• Goals: What success looks like for the agent.
• Guardrails: What the agent must never do.
• Tone: Style and voice expectations.

Guardrails matter

Explicit constraints prevent the agent from hallucinating or oversharing. If the agent should escalate to a human, say so clearly.

Keep it updated

Prompts are living documentation. Update them whenever your product, policies, or workflows change.