MCP Prompt Optimizer

Optimize prompts with research-backed strategies for 15-74% performance improvements.

MCP Prompt Optimizer

Python 3.8+ License: MIT MCP Compatible

A professional-grade MCP (Model Context Protocol) server that provides cutting-edge prompt optimization tools with research-backed strategies delivering 15-74% performance improvements.

✨ Features

🎯 Basic Optimization Strategies

  • Clarity: Simplifies prompts for directness and precision
  • Specificity: Adds detailed constraints and requirements
  • Chain of Thought: Incorporates step-by-step reasoning
  • Few-Shot: Includes example formats for guidance
  • Structured Output: Defines clear output organization
  • Role-Based: Adds expert role context

🚀 Advanced Optimization Strategies

  • Tree of Thoughts (ToT): Multi-path reasoning with 74% success rate on complex tasks
  • Constitutional AI: Self-critique and alignment with safety principles
  • Automatic Prompt Engineer (APE): AI-discovered optimal instruction patterns
  • Meta-Prompting: AI generates its own optimized prompts
  • Self-Refine: Iterative improvement with 20% performance gains
  • TEXTGRAD: Natural language feedback as optimization gradients
  • Medprompt: Multi-technique ensemble achieving 90%+ accuracy
  • PromptWizard: Feedback-driven self-evolving prompts

📋 Professional Domain Templates

Production-ready templates across 11 domains:

  • Business Analysis: Competitive analysis frameworks
  • Product Management: User research synthesis
  • Content Creation: Technical blog posts with SEO optimization
  • Development: Comprehensive code review checklists
  • Communication: Stakeholder updates and project reports
  • Strategy: OKR planning frameworks
  • Operations: Standard Operating Procedures (SOPs)
  • Legal: Contract termination and compliance
  • Customer Experience: Feedback surveys and insights
  • Data Analysis: Data insights and reporting
  • Meeting Management: Effective meeting agendas

🛠️ Installation

Quick Setup

# Clone the repository
git clone <repository-url>
cd mcp-prompt-optimizer

# Create virtual environment (recommended)
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
./install.sh

# Or install manually
pip install -r requirements.txt

# Configure Claude Desktop
python3 setup_interactive.py

Manual Configuration

Add to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "prompt-optimizer": {
      "command": "python3",
      "args": ["/path/to/mcp-prompt-optimizer/prompt_optimizer.py"],
      "env": {}
    }
  }
}

🎮 Usage

Basic Commands

# Analyze prompt quality
"Analyze this prompt: write a blog post about AI"

# Apply specific optimization
"Optimize this prompt using chain_of_thought: explain machine learning"

# Auto-select best strategy
"Auto-optimize: help me debug this code"

# Get domain template
"Get domain template for code_review_checklist"

Advanced Commands

# Use Tree of Thoughts for complex problems
"Apply advanced optimization with tree_of_thoughts: design a microservices architecture"

# Use Constitutional AI for safety-critical tasks
"Apply advanced optimization with constitutional_ai: create content moderation guidelines"

# Use Medprompt for high-accuracy classification
"Apply advanced optimization with medprompt: categorize customer support tickets"

# List available templates
"List all domain templates"

🏗️ Architecture

mcp-prompt-optimizer/
├── prompt_optimizer.py      # Main MCP server
├── advanced_strategies.py   # Research-backed optimization strategies
├── domain_templates.py      # Professional domain templates
├── examples.py              # Usage examples and demonstrations
├── setup_interactive.py     # Automated setup script
└── README.md               # This file

🧪 Testing

# Run basic tests
./test.sh

# Run usage examples
python3 examples.py

📊 Performance Benchmarks

StrategyUse CasePerformance Improvement
Tree of ThoughtsComplex reasoning70-74% success rate
MedpromptClassification tasks90%+ accuracy
Self-RefineIterative improvement20% per iteration
Constitutional AISafety alignmentHigh compliance
Chain of ThoughtStep-by-step tasks15-25% improvement

🔧 Available Tools

Core Tools

  1. analyze_prompt: Analyzes prompt quality and identifies issues
  2. optimize_prompt: Applies specific optimization strategies
  3. auto_optimize: Automatically selects optimal strategy
  4. get_prompt_template: Returns basic templates

Advanced Tools

  1. advanced_optimize: Applies research-backed strategies
  2. get_domain_template: Returns professional domain templates
  3. list_domain_templates: Lists available templates by domain

🎯 Strategy Selection Guide

Prompt TypeRecommended Strategy
Complex problemstree_of_thoughts
Classification tasksmedprompt
Safety-criticalconstitutional_ai
Vague requirementsmeta_prompting
Needs refinementself_refine
General optimizationauto

🤝 Contributing

We welcome contributions! Please:

  1. Fork the repository
  2. Create a feature branch
  3. Add tests for new functionality
  4. Update documentation
  5. Submit a pull request

Adding New Features

  • New Strategy: Add to advanced_strategies.py
  • New Template: Add to domain_templates.py
  • Examples: Add to examples.py

🐛 Troubleshooting

Common Issues

MCP not working?

  • Check Python version: python3 --version (requires 3.8+)
  • Install dependencies: Run ./install.sh or pip install -r requirements.txt
  • Verify MCP installation: pip show mcp
  • Check Claude Desktop logs
  • Restart Claude Desktop

Commands not recognized?

  • Verify configuration file location
  • Check file paths in configuration
  • Run setup script again

Debug Mode

# Test server directly
python3 prompt_optimizer.py

# Verbose logging
export MCP_LOG_LEVEL=debug
python3 prompt_optimizer.py

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

  • Research from Princeton, Google DeepMind, Microsoft Research
  • Anthropic's Constitutional AI framework
  • Stanford's DSPy framework
  • OpenAI's prompt engineering guidelines

📈 Citation

If you use this tool in your research or projects, please cite:

@software{mcp_prompt_optimizer,
  title={MCP Prompt Optimizer: Research-Backed Prompt Optimization for AI Systems},
  author={Bubobot},
  year={2024},
  url={https://github.com/Bubobot-Team/mcp-prompt-optimizer}
}

Built with ❤️ for the AI community

For questions, issues, or contributions, please visit our GitHub repository.

İlgili Sunucular

NotebookLM Web Importer

Web sayfalarını ve YouTube videolarını tek tıkla NotebookLM'e aktarın. 200.000'den fazla kullanıcı tarafından güveniliyor.

Chrome Eklentisini Yükle