AI Prompting Course

Master the Art of Prompt Engineering

Module 4Lesson 2
20 min

The COSTAR Framework

COSTAR (Context, Objective, Style, Tone, Audience, Response) is one of the most structured and adaptable frameworks. Developed by data scientist Sheila Teo, it won Singapore's first GPT-4 Prompt Engineering competition.

Components

  • Context: Provide relevant situational details (customer history, product info, error codes)
  • Objective: Clarify the task with specific goals
  • Style: Specify formatting (paragraphs, bullets, JSON, etc.)
  • Tone: Determine emotional delivery (formal, friendly, empathetic)
  • Audience: Define who receives the response
  • Response: Define output expectations (length, structure, content)

Example

Context: The customer's last package was delayed by 5 days. This is the second complaint in 3 months.
Objective: Write a 100-word apology email acknowledging the delay and offering compensation.
Style: Professional email format
Tone: Formal and apologetic
Audience: Frustrated repeat customer
Response: Must include ticket number, offer 3 solution steps, max 150 words

Exercise

Use the COSTAR framework to create a prompt for a customer service scenario of your choice.