Prompt Engineering Cheatsheet
The ultimate quick-reference guide for crafting perfect prompts. Master ChatGPT, Claude, Gemini, Midjourney, DALL-E, Stable Diffusion, Flux, Kling, Suno, and every major AI tool.
📌 Bookmark This Page • Updated January 2025Text AI Prompts
ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok
🍚 The RICE Framework (Universal Formula)
The most effective structure for text AI prompts across all platforms:
# Complete RICE Example: Role: "You are a senior marketing strategist with 15 years of experience in tech startups." Instructions: "Create a product launch email sequence for a new AI writing tool." Context: "Target audience: content creators and bloggers. Price: $29/month. Key features: SEO optimization, plagiarism checker, tone adjustment." Examples: "Similar to how Grammarly or Jasper positions their products." Format: "Provide 5 emails with subject lines. Each email should be 150-200 words."
🎭 Power Role Starters
Begin prompts with a role for dramatically better results:
"Act as a..." "You are an expert..." "As a professional..." "Imagine you're..." "Take the perspective of..." "You have 20 years of experience as..."
💡 Pro Tip
Add credentials: "You are a Harvard-trained lawyer" or "senior Google engineer with 15 years experience" for more authoritative responses.
📋 Output Format Controls
Specify exactly how you want responses structured:
🧠 Chain-of-Thought Triggers
Force AI to show reasoning for better accuracy on complex tasks:
"Let's think step by step" "Walk me through your reasoning" "Before answering, analyze the problem" "Break this down systematically" "First list the key considerations, then..." "Show your work"
💡 Best For
Math, logic puzzles, complex analysis, debugging code, strategic planning, multi-step problems.
🌡️ Temperature Settings Guide
Control creativity vs. precision (if your tool allows adjustment):
🔒 Constraint Phrases
Set boundaries for more targeted results:
"Do not include..." "Avoid using..." "Only focus on..." "Limit your response to..." "Must include..." "Never mention..." "Use simple language a 10-year-old would understand" "Keep it professional/casual/formal"
⚡ Bad vs. Good Prompts Comparison
"Write about dogs"
"Write a 300-word blog introduction about the mental health benefits of owning a dog for first-time pet owners. Use a warm, conversational tone with 2-3 statistics."
"Help with my resume"
"Act as a senior HR recruiter. Rewrite my software engineer resume bullet points to emphasize impact using the XYZ format (Accomplished X by doing Y, resulting in Z). Focus on quantifiable achievements."
Image AI Prompts
Midjourney, DALL-E, Stable Diffusion, Flux, Leonardo, Ideogram
🎯 Universal Image Prompt Formula
# Complete Example: [Subject] A majestic white wolf [Action] howling at the moon [Environment] on a snow-covered mountain peak [Style] National Geographic wildlife photography [Lighting] moonlit, ethereal blue glow, rim lighting [Camera] Sony A7R IV, 200mm telephoto, f/2.8 [Quality] 8K, ultra detailed, award-winning photo Full prompt: "A majestic white wolf howling at the moon on a snow-covered mountain peak, National Geographic wildlife photography, moonlit scene with ethereal blue glow and rim lighting, shot on Sony A7R IV with 200mm telephoto lens at f/2.8, 8K ultra detailed, award-winning photo"
🎨 Style Keywords
💡 Lighting Keywords
📷 Camera & Quality Keywords
--ar 16:9 # Aspect ratio (16:9, 1:1, 9:16, 4:3, 2:3) --v 6.1 # Version (5.2, 6, 6.1) --style raw # Less stylized, more realistic --stylize 100 # Stylization (0=minimal, 1000=max) --chaos 30 # Variation (0=similar, 100=wild) --no text, blur # Negative prompt (exclude elements) --cref [url] # Character reference (consistent faces) --sref [url] # Style reference (copy style) --q 2 # Quality (0.25, 0.5, 1, 2) --tile # Create seamless patterns
💡 Weight Syntax
Use (keyword:1.3) to emphasize elements. Example: (blue eyes:1.4), (detailed hands:1.2). Higher = stronger effect.
🚫 Universal Negative Prompt (Copy This!)
ugly, deformed, noisy, blurry, distorted, grainy, low quality, worst quality, bad anatomy, bad proportions, extra limbs, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, disfigured, malformed, watermark, text, signature, logo, username, cropped, out of frame, jpeg artifacts, duplicate, morbid, long neck, cross-eyed, extra arms, extra legs, fused fingers, too many fingers, bad hands, missing fingers, error, lowres
⚠️ Note
Midjourney uses --no parameter instead. DALL-E 3 has limited negative prompt support. Flux and SD use full negative prompts.
Video AI Prompts
Kling, Veo 3, MiniMax, Runway, Sora, Pika, Luma
🎯 Video Prompt Formula
# Complete Kling/Runway Example: [Camera] Slow cinematic dolly shot pushing in [Subject] A woman in a flowing red silk dress [Action] walking through a field of lavender, her dress billowing in the wind [Setting] French countryside at golden hour sunset [Style] Shot on ARRI Alexa, anamorphic lens flares, 24fps [Mood] Dreamy, romantic, ethereal atmosphere Full prompt: "Slow cinematic dolly shot pushing in on a woman in a flowing red silk dress walking through a field of lavender, her dress billowing in the wind, French countryside at golden hour sunset, shot on ARRI Alexa with anamorphic lens flares, 24fps, dreamy romantic ethereal atmosphere"
🎥 Camera Movement Keywords
🏃 Motion Description Keywords
Mode: Use "Standard" for speed, "Pro" for quality Input: Start with image-to-video for better control Physics: Describe realistic motion and physics explicitly Duration: Supports up to 2 minutes (extended generation) Audio: No native audio - add with ElevenLabs or Suno later
# Include audio descriptions in your prompt: "[Audio: footsteps on gravel, birds chirping softly, gentle wind]" "[Dialogue: 'Welcome to our store' spoken by a friendly female voice]" "[Music: upbeat electronic background, 120 BPM]" "[SFX: door creaking, clock ticking, rain on window]"
Code AI Prompts
GitHub Copilot, Cursor, Claude Code, ChatGPT, v0
🎯 Code Prompt Formula
# Complete Example: "Write a Python function that: - Takes a list of dictionaries as input - Filters items where 'status' equals 'active' - Sorts by 'created_date' field (newest first) - Returns top 10 results - Handles empty list gracefully (return empty list) - Include type hints and Google-style docstring - Use list comprehension where appropriate"
🐛 Debugging Prompts
"This code should [expected behavior]
but instead it [actual behavior].
Here's the code:
[paste code]
Error message:
[paste error]
What's wrong and how do I fix it?"
💡 Always Include
Expected vs actual behavior, full error message, relevant code context, language/framework versions.
🔍 Code Review Prompts
"Review this code for:
- Performance issues and optimization
- Security vulnerabilities
- Best practice violations
- Edge cases and error handling
- Code readability improvements
- Potential bugs
Provide specific line-by-line feedback."
♻️ Refactoring Prompts
"Refactor this code to:
- Follow SOLID principles
- Extract reusable functions
- Add comprehensive error handling
- Improve variable/function naming
- Reduce complexity (cyclomatic)
- Add logging for debugging
- Make it more testable"
🧪 Test Generation
"Write unit tests for this function using pytest:
- Test normal/happy path cases
- Test edge cases (empty, null, boundary)
- Test error conditions and exceptions
- Use descriptive test names
- Include fixtures where appropriate
- Aim for >90% code coverage"
📝 Documentation Prompts
"Generate comprehensive documentation:
- Function/class docstrings (Google style)
- Parameter descriptions with types
- Return value descriptions
- Raises section for exceptions
- Usage examples
- Any important notes/warnings"
Music AI Prompts
Suno AI, Udio, ElevenLabs, Stable Audio
🎯 Music Prompt Formula
# Complete Suno/Udio Example: [Genre] Indie folk, acoustic [Mood] Nostalgic, warm, emotional [Instruments] Acoustic guitar, soft piano, gentle tambourine [Vocals] Female vocals, soft and breathy, intimate [Tempo] Slow, 80 BPM [Quality] Studio quality, warm analog sound Full prompt: "Indie folk, acoustic guitar, nostalgic and warm, female vocals soft and breathy, gentle piano, light tambourine, slow tempo 80 BPM, campfire feel, studio quality"
🎸 Genre Keywords
😊 Mood Keywords
🎤 Vocal Style Keywords
[Intro] Instrumental guitar intro [Verse 1] Walking down this empty road Memories of days of old The sun sets on what we had But I'm not feeling sad [Chorus] We'll rise again, rise again Like the sun after the rain We'll rise again, rise again [Verse 2] ... [Bridge] [Outro]
✅ Always Do
- Be specific and detailed
- Provide context and examples
- Specify output format clearly
- Set length/scope constraints
- Use step-by-step for complex tasks
- Iterate and refine prompts
❌ Never Do
- Use vague, one-word prompts
- Assume AI knows your context
- Give conflicting instructions
- Expect perfect first results
- Ignore AI's limitations
- Skip verification of outputs
Ready to Create Amazing AI Content?
Put these techniques into practice with our curated prompt collections!
Explore All Prompts