Prompt Engineering Cheatsheet

1. Basic Principles

• Be Specific: Clearly define the task & expectations.
• Use Constraints: Limit responses by length, format, or style.
• Provide Context: Give background for better results.
• Use Examples: Guide with a few-shot approach.
• Iterate & Refine: Modify prompts based on output quality.

---

2. Prompting Techniques

Zero-Shot Prompting

Without examples, the relies only on pre-trained knowledge.

Example:

“ Explain Newton’s Third Law in simple words.”

---

One-Shot Prompting

Provides example to guide the .

Example:

“ Translate ‘Hello’ to French. Answer: Bonjour. Now translate ‘Goodbye’ to French.”

---

Few-Shot Prompting

Uses examples for accuracy.

Example:

“ Translate:

• ‘Hello’ → ‘Bonjour’
• ‘Goodbye’ → ‘Au revoir’
• ‘Thank you’ → ?”

Response: “Merci.”

---

Chain-of-Thought (CoT) Prompting

Encourages step-by-step reasoning.

Example:

“ A train moves at 60 km/h for 2.5 hours. How far does it travel? Explain step by step.”

Response:

Step : Speed = 60 km/h
Step : Time = 2.5
Step : Distance = Speed × Time = 60 × 2.5 = 150 km.

---

Instruction-Based Prompting

Directly instructs with clear guidelines.

Example:

“ Summarize this article in bullet points.”

---

Role-Based Prompting

Assigns a persona for understanding.

Example:

“ You are a historian. Explain the impact of the Chola dynasty in South India.”

---

Delimiter Prompting

Keeps focused by using delimiters.

Example:

“ Summarize the following text between triple quotes:
”’ The mitochondria is the powerhouse of the cell…”'”

---

3. Advanced Prompting Techniques

Prompt Chaining

Breaks down complex tasks into sequential prompts.

Example:
Generate an outline for an essay. Expand each section into paragraphs. Refine grammar & style.

---

Self-Consistency

Runs multiple variations of a prompt & picks best output.

Example:

“ Solve 27 × 43. Explain your reasoning.”
(AI runs multiple approaches & picks the most consistent result.)

---

Retrieval-Augmented Generation (RAG)

Fetches relevant external data before generating a response.

Example:

“ Summarize the latest research on quantum computing using external sources.”

---

4. Common Issues & Fixes

Issue	Fix
gives vague answers	Use specific instructions & constraints
Response is too long	Specify word/character limit
misunderstands context	Provide clear background
Output lacks depth	Use chain-of-thought or role-based prompts
Hallucinated facts	Use RAG-based or fact-checking prompts

---

5. Applications of Prompt Engineering

• Chatbots & Assistants – Personalized AI interactions.
• Content Creation – Blogs, articles, & creative writing.
• Education – AI tutors, explanations, & summaries.
• Coding – Generating & debugging code.
• Data Analysis – Summarizing reports & extracting insights.

By mastering these techniques, you can create highly effective prompts for AI-driven tasks & maximize the accuracy of AI-generated outputs.