Role prompts and constraints

Get better answers by assigning roles and boundaries

---

Learning goal

• Use roles to guide style and depth.
• Add constraints to control scope.

---

Why it matters

• Role prompts produce domain-appropriate language.
• Constraints prevent overlong or off-topic answers.

---

Explanation

• Role examples: “You are a tech writer / teacher / reviewer.”
• Constraints: word count, bullet count, audience, exclusions.
• Safety: “If unsure, ask for missing info; do not fabricate data.”

---

Examples

• Good: “Act as a senior UX writer. Rewrite this microcopy for beginners. Limit to 3 bullets, plain language.”
• Poor: “Rewrite this.”

---

Guided exercise (10–15 min)

1. Pick a task: add a role and 2 constraints.
2. Generate output; refine constraints if too long/off-topic.

Independent exercise (5–10 min)

Apply a different role (e.g., teacher vs reviewer) to the same task; compare outputs.

---

Self-check

• Role stated clearly.
• Constraints include length/audience/exclusions.
• Safety clause included.

---

Optional deepening

• Role prompting ideas: https://platform.openai.com/docs/guides/prompt-engineering