Anatomy of a Well-Crafted Prompt
Click each block to learn what it does. A strong prompt is built from layered components — each one sharpens the model's output.
Prompt Builder — click a block to inspect 👋 Click any block on the left
Each colour-coded block represents a different component of an effective prompt. Select one to see a detailed explanation, common mistakes, and tips for improvement.
🧩 The Six Blocks
A complete prompt typically contains:
Role — Who is the AI?
Context — What background info?
Task — What should it do?
Format — How should it respond?
Examples — Show don't just tell
Constraints — What to avoid?
📐 Token Economy
Every word costs tokens. Good prompts are specific, not verbose. Aim for clarity over length — a 50-token prompt with a clear role + task often outperforms a 300-token vague essay.
Core Techniques at a Glance
🔗Chain of Thought
Ask the model to reason step-by-step before answering. Dramatically improves accuracy on logic, math, and multi-step tasks.
📸Few-Shot Examples
Provide 2–5 input/output examples before your actual request. Shows the model the exact pattern you want.
🎭Role Prompting
"You are an expert X…" primes the model to adopt a persona, vocabulary, and depth of knowledge appropriate to that role.
📋Output Formatting
Specify structure: JSON, markdown, numbered list, table. Structured output is easier to parse and use programmatically.
🔄Self-Consistency
Generate multiple responses and pick the most common answer. Reduces variance on reasoning tasks with a single prompt.
🌲Tree of Thought
Explore multiple reasoning paths simultaneously. The model evaluates branches and backtracks — better for complex planning.
Before & After — Real Prompt Comparisons
The same request, rewritten with proper prompt engineering. The difference in output quality is dramatic.
Prompt Engineering Techniques