Prompt Engineering Plus#

Enhanced prompt engineering with model-specific optimization, advanced techniques, and evaluation metrics.

Features#

• Model-Specific Tips: Optimized prompts for Claude, GPT-4, Gemini, FLUX, Veo
• Advanced Techniques: Chain-of-thought, few-shot, system prompts
• Prompt Templates: Ready-to-use templates for common tasks
• Evaluation Metrics: Measure prompt effectiveness
• Optimization Workflow: Iterative improvement process

Quick Reference#

TaskTechniqueTemplateCode reviewRole + constraintsReview this [language] code for...Content writingAudience + toneWrite a [type] about [topic] for [audience]Image generationSubject + style + composition[Subject], [style], [lighting], [quality]Data extractionSchema + examplesExtract fields matching this schema...AnalysisStep-by-stepAnalyze [data] by [method]

LLM Prompting#

Basic Structure#

text

[Role/Context] + [Task] + [Constraints] + [Output Format]

Role Prompting#

python

# Expert role
"You are a senior software engineer with 15 years of experience in [domain]."

# Teaching role
"You are a patient teacher explaining [topic] to a beginner."

# Critic role
"You are a constructive critic reviewing [work] with specific feedback."

Chain-of-Thought#

python

# Explicit steps
"Think through this step by step before giving the final answer."

# Structured reasoning
"1. First, identify the key facts...
2. Then, analyze the relationships...
3. Finally, draw conclusions..."

Few-Shot Examples#

python

# Example format
"Convert these sentences to formal business English:

Example 1:
Input: gonna send u the report tmrw
Output: I will send you the report tomorrow.

Example 2:
Input: cant make the meeting, something came up
Output: I apologize, but I will be unable to attend the meeting due to an unforeseen circumstance.

Now convert:
Input: hey can we push the deadline back a bit?"

Output Format Specification#

python

# JSON schema
"Return a JSON array with objects containing 'text', 'sentiment' (positive/negative/neutral), and 'confidence' (0-1)."

# Structured output
"For each issue found, provide:
- Line number
- Issue description
- Severity (high/medium/low)
- Suggested fix"

Constraint Setting#

python

# Word limits
"Summarize in exactly 3 bullet points, each under 20 words."

# Format constraints
"Return only valid JSON, no explanation."

# Content constraints
"Focus only on actionable insights, not background information."

Image Generation Prompting#

Basic Structure#

text

[Subject] + [Style] + [Composition] + [Lighting] + [Technical]

Style Keywords#

python

# Photography
"shot on Kodak Portra 400 film, soft natural lighting, shallow depth of field"

# Digital art
"digital painting, concept art, artstation, trending"

# Oil painting
"oil painting, masterpiece, classical, detailed brushwork"

# 3D render
"3D render, octane render, cinema 4D, detailed"

Quality Keywords#

python

# High quality
"photorealistic, 8K, ultra detailed, sharp focus, professional, masterpiece"

# Artistic
"intricate details, award-winning, best quality"

# Technical
"high resolution, 4K, detailed texture, sharp"

Negative Prompts#

python

# Common negatives
"blurry, distorted, extra limbs, watermark, text, low quality, cartoon, anime"

# Photography negatives
"overexposed, underexposed, grainy, noisy, artifacts"

# Art negatives
"ugly, deformed, disfigured, poor proportions, bad anatomy"

Video Prompting#

Basic Structure#

text

[Shot Type] + [Subject] + [Action] + [Setting] + [Style]

Camera Movement#

python

# Tracking
"Slow tracking shot following [subject] through [setting]"

# Pan
"Smooth pan across [scene] revealing [subject]"

# Zoom
"Slow zoom into [subject] with [effect]"

# Handheld
"Handheld camera following [subject] with natural movement"

Temporal Keywords#

python

# Speed
"slow motion, timelapse, real-time, smooth motion"

# Duration
"continuous shot, quick cuts, frozen moment"

# Transitions
"dissolve, fade, wipe, morph"

Model-Specific Tips#

Claude#

• Excels at nuanced instructions
• Responds well to role-playing
• Good at following complex constraints
• Prefers explicit output formats

GPT-4#

• Strong at code generation
• Works well with examples
• Good structured output
• Responds to "let's think step by step"

Gemini#

• Good at multi-modal tasks
• Works well with visual prompts
• Strong at analysis tasks
• Prefers clear instructions

FLUX#

• Detailed subject descriptions
• Style references work well
• Lighting keywords important
• Negative prompts supported

Veo#

• Camera movement keywords
• Cinematic language works well
• Action descriptions important
• Include temporal context

Evaluation Metrics#

Effectiveness Score#

MetricDescriptionMeasurementClarityHow clear is the prompt?1-5 ratingSpecificityHow specific are the instructions?Word count, detail levelConsistencyHow consistent are results?Variance across runsQualityHow good is the output?Manual review

Prompt Quality Checklist#

• Clear objective stated
• Constraints defined
• Output format specified
• Examples provided (if needed)
• No conflicting instructions
• Appropriate length

Optimization Workflow#

1. Start Simple#

python

# Basic prompt
"Write a summary of this article."

2. Add Specificity#

python

# More specific
"Write a 3-sentence summary of this article, focusing on key findings."

3. Add Constraints#

python

# With constraints
"Write a 3-sentence summary of this article, focusing on key findings. Use formal business English. Include specific statistics if mentioned."

4. Add Examples#

python

# With examples
"Write a 3-sentence summary of this article, focusing on key findings. Use formal business English. Include specific statistics if mentioned.

Example:
Input: 'Study finds 85% of users prefer interface X over Y.'
Output: 'A recent study reveals that 85% of users demonstrate a preference for Interface X compared to Interface Y. The findings suggest significant usability advantages for Interface X. This data should inform future interface design decisions.'"

5. Iterate Based on Results#

python

# Refined prompt
"Write a 3-sentence summary of this article, focusing on key findings. Use formal business English. Include specific statistics if mentioned. Avoid technical jargon. Start with the most important finding.

Example:
Input: 'Study finds 85% of users prefer interface X over Y.'
Output: 'A recent study reveals that 85% of users demonstrate a preference for Interface X compared to Interface Y. The findings suggest significant usability advantages for Interface X. This data should inform future interface design decisions.'"

Common Mistakes#

MistakeProblemFixToo vagueUnpredictable outputAdd specificsToo longModel loses focusPrioritize key infoConflictingConfuses modelRemove contradictionsNo formatInconsistent outputSpecify formatNo examplesUnclear expectationsAdd few-shot

Prompt Templates#

Code Review#

python

"Review this [language] code for:
1. Bugs and logic errors
2. Security vulnerabilities
3. Performance issues
4. Code style/best practices

Code:
[code]

For each issue found, provide:
- Line number
- Issue description
- Severity (high/medium/low)
- Suggested fix"

Content Writing#

python

"Write a [content type] about [topic].

Audience: [target audience]
Tone: [formal/casual/professional]
Length: [word count]
Key points to cover:
1. [point 1]
2. [point 2]
3. [point 3]

Include: [specific elements]
Avoid: [things to exclude]"

Data Analysis#

python

"Analyze this data:
[data]

Provide:
1. Key insights (3-5 findings)
2. Patterns or trends
3. Anomalies or outliers
4. Recommendations based on findings

Use statistical terms where appropriate. Include confidence levels for key claims."

Image Generation#

python

"[Subject with details], [setting/background], [lighting type],
[art style or photography style], [composition], [quality keywords]"

Best Practices#

1. Start simple - Begin with basic prompts, add complexity as needed
2. Be specific - Vague prompts produce vague results
3. Use examples - Few-shot learning improves consistency
4. Set constraints - Word limits, formats, tone
5. Iterate - Refine based on results
6. Evaluate - Measure effectiveness
7. Document - Save successful prompts