A powerful Model Context Protocol (MCP) server that helps you create, validate, manage, and optimize prompts using the RISEN framework.
What is RISEN?
RISEN is a structured prompt engineering framework with 5 components:
- Role: Define the AI's persona/expertise
- Instructions: Clear directives for the task
- Steps: Breakdown of the process
- Expectations: Desired outcome/format
- Narrowing: Constraints or creative elements
Features
🎯 Core Functionality
- Template Management: Create, store, and organize RISEN prompt templates
- Variable Support: Use
{{variables}} for dynamic, reusable prompts - Validation Engine: Real-time structure checking and quality scoring
- Performance Tracking: Monitor prompt effectiveness with ratings and analytics
- AI Suggestions: Get improvement recommendations based on best practices
🚀 Advanced Features
- A/B Testing: Compare different prompt variations
- Cross-AI Integration: Works with your Cross-AI tool to test prompts on multiple models
- Knowledge Base Integration: Save successful prompts for future reference
- Natural Language Conversion: Transform regular requests into RISEN format
- Template Library: Pre-built templates for common tasks
Installation
- Clone the repository:
git clone https://github.com/YOUR_USERNAME/mcp-risen-prompts.git
cd mcp-risen-prompts
- Install dependencies:
- Test the server:
- The server is now ready to be configured in Claude Desktop
Configuration
Add to your Claude Desktop config:
macOS
~/Library/Application Support/Claude/claude_desktop_config.json
Windows
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"risen-prompts": {
"command": "node",
"args": ["/path/to/mcp-risen-prompts/server.js"]
}
}
}
Usage Examples
Creating a Template
Use risen_create to make a new template:
- Name: "Code Review"
- Role: "Senior software engineer with 15+ years experience"
- Instructions: "Review the provided code for quality and security"
- Steps: ["Analyze structure", "Check for bugs", "Suggest improvements"]
- Expectations: "Detailed line-by-line feedback with examples"
- Narrowing: "Focus on critical issues first"
Executing a Template
Use risen_execute with variables:
- Template ID: [your-template-id]
- Variables: {"language": "Python", "framework": "Django"}
After using a prompt, track its effectiveness:
- Use risen_track
- Rate 1-5 stars
- Add notes about what worked/didn't work
- risen_create - Create new RISEN templates
- risen_validate - Check structure and get suggestions
- risen_execute - Run templates with variables
- risen_track - Record performance metrics
- risen_search - Find templates by tags/rating
- risen_analyze - Get insights on template performance
- risen_suggest - AI-powered improvement recommendations
- risen_convert - Transform natural language to RISEN
Template Examples
Blog Post Writer
Role: Content strategist and SEO expert
Instructions: Write an engaging blog post about {{topic}}
Steps:
1. Research keywords and trends
2. Create compelling headline
3. Develop main points with examples
4. Include statistics and sources
5. Write conclusion with CTA
Expectations: 1500-2000 words, SEO-optimized, engaging tone
Narrowing: Use conversational tone, include 3-5 keywords naturally
Data Analysis
Role: Data scientist specializing in {{domain}}
Instructions: Analyze {{dataset}} to uncover insights
Steps:
1. Perform exploratory data analysis
2. Identify key trends and patterns
3. Run statistical tests
4. Create visualizations
5. Provide recommendations
Expectations: Clear insights with statistical backing
Narrowing: Focus on {{specific_metrics}} and business impact
Quality Scoring
Templates are scored out of 100 based on:
- Role specificity (20 points)
- Instruction clarity (20 points)
- Step detail (20 points)
- Expectation metrics (20 points)
- Narrowing focus (20 points)
Best Practices
- Be Specific: Vague roles like "assistant" score lower than "Senior Python developer with AWS expertise"
- Use Variables: Make templates reusable with
{{variables}} - Measurable Expectations: Include numbers (word count, examples needed, etc.)
- Clear Steps: Each step should be actionable and specific
- Test & Iterate: Use tracking to refine templates over time
Execute the same RISEN prompt across multiple AI models:
- Create/select a RISEN template
- Use Cross-AI to run it on ChatGPT, Gemini, and Claude
- Compare results and track which model performs best
With Knowledge Base
Save successful prompts for future reference:
- Create and test a RISEN prompt
- Once proven effective, save to Knowledge Base
- Search and retrieve proven prompts by topic
Troubleshooting
Template not validating?
- Ensure all required fields are filled
- Check that steps is an array, not a string
- Verify variables are properly declared
Variables not replacing?
- Use exact syntax:
{{variable_name}} - Ensure variable names match in declaration and usage
- Check that all variables have values when executing
Low quality scores?
- Add more detail to each component
- Include specific metrics in expectations
- Use domain-specific language in role
Future Roadmap
Contributing
Found a bug or have a feature request? Feel free to contribute!
License
MIT License - feel free to use and modify as needed.