Aim: To control an LLM response using explicit output structure together with positive and negative constraints.
Prompt / Activity:
PROMPT:

Create a revision note on "Prompt Injection" for postgraduate students.

Positive requirements:
- Start with a one-sentence definition.
- Include exactly 4 warning signs.
- Include a table with columns: Attack Pattern, Example, Mitigation.
- End with 3 defensive design rules.

Negative constraints:
- Do not provide instructions for bypassing safeguards.
- Do not include executable attack payloads.
- Do not exceed 300 words.
- Do not add any section other than those requested.

Use clear academic language.
