Prompting Pattern Library

Comprehensive collection of 25+ proven AI prompting patterns with model-specific guidance, failure mode analysis, and orchestration strategies for advanced AI interactions and workflow optimization.

Essential for anyone working with AI models who wants to move beyond basic prompting to develop sophisticated, reliable AI interaction patterns that consistently produce high-quality results.

Skill Structure

This skill is part of Nate's Substack Skills collection:

Main Files:

SKILL.md - Complete prompting pattern library
references/prompt-patterns.md - Detailed pattern catalog
references/failure-modes.md - Common failure analysis
references/model-quirks.md - Model-specific behaviors
references/orchestration-patterns.md - Multi-agent workflows

Full Collection: Nate's Substack Skills - Explore all skills!

Core Philosophy

Pattern-Based Prompting Over Ad-Hoc Approaches

Effective AI interaction relies on proven patterns rather than improvised prompts:

Reusable Templates: Standardized approaches for common tasks
Predictable Outcomes: Consistent results through structured prompting
Model Optimization: Patterns tailored to specific AI model strengths
Failure Prevention: Understanding and avoiding common prompt failures

Prompting Success Factors

Effective Patterns:

Clear role and context definition
Specific output format requirements
Step-by-step reasoning guidance
Relevant examples and demonstrations
Error handling and edge case coverage

Common Failures:

Vague or ambiguous instructions
Missing context or constraints
Overwhelming complexity in single prompts
Inconsistent formatting expectations
Lack of validation or feedback loops

Pattern Categories

Foundation Patterns

Role Prompting:

You are a [specific role] with [relevant expertise].
Your task is to [specific objective].
Consider [relevant constraints/context].

Chain-of-Thought:

Let's work through this step by step:
1. First, [initial analysis]
2. Then, [next consideration]
3. Finally, [conclusion/recommendation]

Few-Shot Learning:

Here are examples of the desired output:

Example 1: [input] → [desired output]
Example 2: [input] → [desired output]

Now apply this pattern to: [new input]

2. Output Control Patterns

Structured Output:

Please provide your response in the following format:
- Summary: [brief overview]
- Analysis: [detailed examination]
- Recommendations: [actionable next steps]
- Confidence: [level of certainty]

Format Enforcement:

Output your response as valid JSON with these keys:
{
  "reasoning": "step-by-step thinking",
  "conclusion": "final answer",
  "confidence": "high/medium/low"
}

Length Control:

Provide a [specific length] response that:
- Covers [key points]
- Excludes [unnecessary details]
- Uses [appropriate tone/style]

3. Reasoning Enhancement Patterns

Analytical Patterns

Critical Thinking Framework:

Analyze this by considering:
1. What assumptions are being made?
2. What evidence supports/contradicts this?
3. What are alternative explanations?
4. What are the implications if true/false?

Pro/Con Analysis:

Evaluate [topic] by:
1. Listing 3-5 key advantages
2. Identifying 3-5 significant disadvantages  
3. Weighing the overall balance
4. Recommending a position with rationale

Scenario Planning:

Consider these scenarios for [situation]:
- Best case: [optimistic outcome and likelihood]
- Most likely: [realistic outcome and likelihood]
- Worst case: [pessimistic outcome and likelihood]
Plan responses for each scenario.

4. Context Management Patterns

Context Priming:

Given this background context: [relevant information]
And these current constraints: [limitations]
Help me [specific request] while considering [key factors].

Perspective Taking:

Approach this from the perspective of:
- [Stakeholder 1]: Their concerns would be [specific interests]
- [Stakeholder 2]: They would prioritize [different interests]
- [Stakeholder 3]: Their focus would be [third perspective]

Domain Expertise:

As an expert in [specific field], with knowledge of:
- [Key concept 1]
- [Key concept 2]
- [Key concept 3]
Apply this expertise to [specific problem].

Model-Specific Optimization

Claude Strengths

Structured Reasoning: Claude excels with clear analytical frameworks:

Please analyze this systematically:
1. Break down the core components
2. Examine relationships between elements
3. Identify potential issues or opportunities
4. Synthesize findings into actionable insights

Ethical Considerations:

Consider the ethical implications of [scenario]:
- Stakeholder impact analysis
- Potential unintended consequences
- Alignment with ethical principles
- Recommendations for responsible action

Complex Document Analysis:

Analyze this document by:
1. Summarizing key themes
2. Identifying supporting evidence
3. Noting any gaps or inconsistencies
4. Providing strategic recommendations

GPT-4 Patterns

Creative Problem Solving:

Generate creative solutions for [problem] by:
1. Reframing the problem from different angles
2. Drawing analogies from other domains
3. Combining seemingly unrelated concepts
4. Proposing unconventional approaches

Step-by-Step Instructions:

Create detailed instructions for [task]:
1. Prerequisites and preparation
2. Step-by-step execution
3. Quality checkpoints
4. Troubleshooting common issues

Gemini Patterns

Multimodal Analysis:

Analyze this [image/document/data] by:
1. Describing visual/structural elements
2. Extracting key information
3. Identifying patterns or anomalies
4. Connecting to broader context

Advanced Orchestration Patterns

Agent Coordination

Sequential Processing:

Agent 1: Research and gather information on [topic]
Agent 2: Analyze findings and identify key insights
Agent 3: Generate recommendations based on analysis
Agent 4: Review and refine final output

Parallel Analysis:

Simultaneously analyze [problem] from these angles:
- Technical feasibility (Agent A)
- Business viability (Agent B)  
- User experience impact (Agent C)
- Risk assessment (Agent D)
Synthesize findings for comprehensive view.

Iterative Refinement:

Round 1: Generate initial [output type]
Round 2: Critique and identify improvements
Round 3: Refine based on feedback
Round 4: Final quality check and validation

Validation Patterns

Self-Correction:

After providing your initial response:
1. Review for accuracy and completeness
2. Identify potential errors or gaps
3. Correct any issues found
4. Confirm final answer quality

Cross-Validation:

Verify this conclusion by:
1. Checking against established principles
2. Testing with alternative scenarios
3. Considering counter-arguments
4. Confirming logical consistency

Failure Mode Prevention

Failure Analysis

Ambiguity Issues:

Problem: Vague instructions leading to inconsistent outputs
Solution: Specific requirements with clear examples
Pattern: Use explicit constraints and format specifications

Context Overload:

Problem: Too much information overwhelming the model
Solution: Prioritize and structure information hierarchically
Pattern: Lead with most important context, supporting details after

Scope Creep:

Problem: Requests expanding beyond intended boundaries
Solution: Clear boundaries and explicit limitations
Pattern: Define what to include AND what to exclude

Error Prevention Strategies

Input Validation:

Before proceeding, confirm you understand:
1. The specific task requirements
2. Expected output format
3. Key constraints or limitations
4. Success criteria for the response

Boundary Setting:

Focus specifically on [defined scope].
Do not include [excluded elements].
If unclear about boundaries, ask for clarification.

Quality Gates:

Before finalizing your response:
- Does it directly address the request?
- Is the format correct and complete?
- Are claims supported by evidence?
- Would this be helpful to the user?

Specialized Application Patterns

Content Patterns

Writing Framework:

Create [content type] that:
1. Targets [specific audience]
2. Achieves [specific goal]
3. Follows [style/tone guidelines]
4. Includes [required elements]
5. Avoids [common pitfalls]

Editing and Refinement:

Improve this content by:
1. Enhancing clarity and readability
2. Strengthening key arguments
3. Improving flow and structure
4. Ensuring consistency in tone
5. Correcting any errors

Analysis and Research

Research Framework:

Research [topic] by:
1. Defining key research questions
2. Identifying relevant sources and methodologies
3. Analyzing findings systematically
4. Drawing evidence-based conclusions
5. Noting limitations and future directions

Data Analysis Pattern:

Analyze this data by:
1. Describing the dataset characteristics
2. Identifying patterns and trends
3. Testing relevant hypotheses
4. Quantifying relationships
5. Interpreting business implications

Problem Solving

Systematic Problem Solving:

Address [problem] using this framework:
1. Problem definition and scope
2. Root cause analysis
3. Solution generation and evaluation
4. Implementation planning
5. Success metrics and monitoring

Decision Support:

Help me decide [decision] by:
1. Clarifying decision criteria
2. Evaluating available options
3. Assessing risks and benefits
4. Recommending optimal choice
5. Planning implementation steps

Quality Assurance Patterns

Quality Control

Completeness Check:

Ensure your response includes:
- Direct answer to the question
- Supporting reasoning or evidence
- Acknowledgment of limitations
- Clear next steps or recommendations

Accuracy Verification:

Validate your response by:
1. Checking facts against reliable sources
2. Confirming logical consistency
3. Testing conclusions with examples
4. Identifying potential errors or gaps

Relevance Assessment:

Confirm your response:
- Directly addresses the user's needs
- Stays within the requested scope
- Provides actionable information
- Matches the expected format

Iterative Improvement

Feedback Integration:

Based on feedback that [specific feedback]:
1. Identify areas for improvement
2. Revise problematic sections
3. Enhance overall quality
4. Verify improvements address concerns

Refinement Process:

Improve this output by:
1. Analyzing current strengths and weaknesses
2. Identifying specific enhancement opportunities
3. Implementing targeted improvements
4. Validating enhanced version quality

Implementation Guidelines

Pattern Selection

Task-Based Selection:

Analysis tasks: Use reasoning enhancement patterns
Creative tasks: Apply generation and ideation patterns
Technical tasks: Employ structured output patterns
Research tasks: Utilize investigation and validation patterns

Model-Based Optimization:

Claude: Structured reasoning and ethical analysis
GPT-4: Creative problem-solving and detailed instructions
Gemini: Multimodal analysis and visual processing

Adaptation Strategies

Context Adaptation:

Modify examples for domain relevance
Adjust language for audience expertise
Include industry-specific considerations
Reference relevant frameworks or standards

Output Optimization:

Specify format requirements clearly
Define quality criteria explicitly
Include validation checkpoints
Plan for iterative refinement

Measurement and Optimization

Pattern Effectiveness

Success Metrics:

Output quality and relevance
Consistency across iterations
Time to achieve desired results
User satisfaction with responses

Performance Tracking:

Pattern usage frequency
Success rate by pattern type
Common failure modes
Improvement opportunities

Continuous Improvement

Pattern Evolution:

Regular review of pattern effectiveness
Integration of new model capabilities
User feedback incorporation
Best practice documentation

Library Maintenance:

Pattern catalog updates
Failure mode analysis
Model-specific optimization
Community contribution integration

About This Skill

This skill was created by Nate Jones as part of his comprehensive Nate's Substack Skills collection. Learn more about Nate's work at Nate's Newsletter.

Explore the full collection to discover all 10+ skills designed to enhance your Claude workflows!

Comprehensive prompting pattern library that transforms basic AI interactions into sophisticated, reliable workflows through proven templates, model-specific optimization, and systematic quality assurance.