Promptheus

# Promptheus: Sample Prompts This file contains sample prompts to test and demonstrate Promptheus capabilities, particularly the adaptive interaction model that detects task types (analysis vs generation). ## How to Use These Prompts Promptheus intelligently detects task types: - **Analysis tasks** (research, exploration, debugging): Skip questions by default - **Generation tasks** (writing, creating content): Offer clarifying questions Each prompt below indicates the expected behavior for testing purposes. --- ## Generation Tasks (Triggers Clarifying Questions) These prompts should trigger the question-answer workflow because they involve creating new content. ### 1. Futuristic Story ``` Write a short, futuristic story for a young adult audience about a teenager who discovers an AI that can dream. The story should have a hopeful tone and be around 500 words. ``` **Expected**: Asks about protagonist details, dream themes, conflict type, ending preference ### 2. Marketing Campaign Slogan ``` Generate five catchy slogans for a new brand of eco-friendly coffee. ``` **Expected**: Asks about brand personality, target audience, key differentiators, tone ### 3. Ambiguous Prompt ``` Social media post about AI. ``` **Expected**: Asks about platform, audience, tone, length, key message ### 4. Poetry with Constraints ``` Write a haiku about a bustling city at night. ``` **Expected**: May ask about specific city, mood, imagery preferences ### 5. Screenplay Dialogue ``` Write a tense dialogue scene between a detective and a suspect who knows more than they are letting on. The scene should be no more than 200 words. ``` **Expected**: Asks about setting, character backgrounds, revelation level ### 6. Blog Post Outline ``` Create an outline for a technical blog post about microservices architecture. ``` **Expected**: Asks about target audience, depth level, specific topics, post length ### 7. Python Function with Requirements ``` Create a Python function named calculate_ema that calculates the Exponential Moving Average of a list of numbers. The function should take two arguments: a list of prices and a span (integer). It must include type hints, a docstring explaining its usage, and handle potential errors like an empty input list. ``` **Expected**: Asks about error handling strategy, return type format, validation requirements ### 8. Product Description ``` Write a compelling product description for a smart home security system targeting tech-savvy homeowners. ``` **Expected**: Asks about key features, price range, competitive advantages, tone ### 9. Newsletter Content ``` Create engaging content for a weekly tech newsletter covering recent AI developments. ``` **Expected**: Asks about audience level, content format, specific topics, length --- ## Analysis Tasks (Skip Questions by Default) These prompts should skip the question phase and directly enhance/analyze because they're about understanding, debugging, or explanation. ### 10. Debug JavaScript Code ``` My JavaScript code isn't working as expected. When I click the button, the text content doesn't update. Can you tell me why? <p id='my-text'>Hello World</p> <button id='my-button'>Change Text</button> const button = document.getElementById('my-button'); button.addEventListener('click', () => { const text = document.getElementById('my-text'); text.innerHtml = 'Hello, New World!'; }); ``` **Expected**: Skip questions, directly identifies the typo (`innerHtml` vs `innerHTML`) ### 11. Performance Analysis ``` I have a Python script that is running slower than expected. Analyze this code for performance bottlenecks and suggest specific optimizations. def process_data(data_list): results = [] for item in data_list: processed_item = item * item time.sleep(0.1) results.append(processed_item) return results ``` **Expected**: Skip questions, analyzes bottlenecks (sleep, list append), suggests optimizations ### 12. Git Command Explanation ``` Explain the git rebase -i HEAD~3 command. What does it do, and what are some common use cases for it in a development workflow? ``` **Expected**: Skip questions, provides direct explanation with examples ### 13. Concept Explanation ``` Explain the concept of blockchain to someone with no technical background. ``` **Expected**: Skip questions, provides simplified explanation with analogies ### 14. Architecture Comparison ``` Compare and contrast the architectural patterns of microservices and a monolithic backend. Focus on scalability, development speed, and operational complexity. ``` **Expected**: Skip questions, provides direct comparative analysis ### 15. Customer Feedback Analysis ``` Analyze the following customer reviews for a fictional product and identify the top 3 most common complaints and top 2 most praised features. Review 1: "The battery life is amazing! But the screen is too dim." Review 2: "I love how fast it is, but it keeps crashing." Review 3: "Super fast and reliable. Best purchase ever." Review 4: "The screen is hard to see outdoors. Battery is great though." ``` **Expected**: Skip questions, directly provides sentiment analysis and categorization ### 16. Research Summary ``` Summarize the key findings from recent research papers on transformer architectures in natural language processing. ``` **Expected**: Skip questions, provides direct summary (or may ask for specific papers if too broad) ### 17. SQL Query Analysis ``` Write a SQL query to find all users who have signed up in the last 30 days and made at least one purchase. Assume the tables are users (with id, name, signup_date) and purchases (with id, user_id, purchase_date, amount). ``` **Expected**: May skip questions or ask minimal clarifications about database engine/performance needs --- ## Edge Cases & Testing Scenarios ### 18. Minimal Context (Should Trigger Questions) ``` Write something interesting. ``` **Expected**: Asks about topic, format, audience, length, purpose ### 19. Technical with Context (May Skip) ``` Create a Dockerfile for a Node.js application that uses Express and PostgreSQL. ``` **Expected**: May ask about Node version, optimization needs, or skip and provide standard solution ### 20. Code Refactoring (Analysis) ``` Refactor this nested callback code into async/await syntax while maintaining error handling. ``` **Expected**: Skip questions, provides direct refactoring ### 21. Business Writing (Generation) ``` Draft a polite and concise email to a colleague asking for an update on a project that is past its deadline. ``` **Expected**: Asks about relationship with colleague, urgency level, project context ### 22. Study Guide (Generation) ``` Create a study guide for learning Python programming from scratch over 4 weeks. ``` **Expected**: Asks about time commitment, learning goals, resources available, prior experience --- ## Testing Skip Mode Flag Use the `-s` or `--skip-questions` flag to force skipping questions regardless of task type. ### 23. Force Skip on Generation Task ``` promptheus -s "Write a blog post about async programming" ``` **Expected**: Skips questions, directly enhances the prompt ### 24. Force Skip on Analysis Task ``` promptheus -s "Explain Docker containers" ``` **Expected**: Skips questions (consistent with normal behavior for analysis tasks) --- ## Testing Refine Mode Flag Use the `-r` flag to force clarifying questions even for analysis tasks. ### 25. Force Questions on Analysis Task ``` promptheus -r "Explain recursion" ``` **Expected**: Forces questions despite being an analysis task --- ## Multi-Modal Task Detection These prompts blend analysis and generation to test classification accuracy. ### 26. Analysis + Generation ``` Review this function for security issues and rewrite it to be more secure. ``` **Expected**: May ask about security requirements, context, or skip for direct analysis+fix ### 27. Tutorial Creation (Generation) ``` Create a beginner-friendly tutorial explaining how HTTP requests work, including GET and POST methods. ``` **Expected**: Asks about prior knowledge assumptions, depth, examples needed, format ### 28. Documentation (Generation) ``` Write comprehensive API documentation for a user management service with authentication endpoints. ``` **Expected**: Asks about API structure, authentication method, example requirements --- ## Testing Notes - **Task classification** is the key feature being tested - **Generation tasks** should ask clarifying questions to gather requirements - **Analysis tasks** should skip questions and provide direct analysis/explanation - **Edge cases** test the boundary between task types - Use **skip mode** (`-s`) when you want to bypass questions for any task - Use **refine mode** (`-r`) to force questions on any task --- ## MCP Server Testing Examples These examples demonstrate how to test the Promptheus MCP server integration with MCP-compatible clients. ### Basic MCP Tool Testing **Test 1: List Available Providers** ```bash # MCP client call mcp-client call list_providers # Expected response format { "type": "success", "providers": { "google": {"configured": true, "model": "gemini-2.0-flash-exp"}, "openai": {"configured": false, "error": "No API key found"} } } ``` **Test 2: Validate Environment** ```bash # MCP client call with connection test mcp-client call validate_environment --test-connection true # Expected response shows provider status { "type": "success", "validation": { "google": {"configured": true, "connection_test": "passed"} } } ``` ### MCP Refinement Workflow Testing **Test 3: Direct Refinement (No Questions)** ```bash # MCP client call for analysis task mcp-client call refine_prompt --prompt "Explain Docker containers" # Expected: Direct refinement without questions { "type": "refined", "prompt": "Provide a comprehensive explanation of Docker containers...", "next_action": "This refined prompt is now ready to use..." } ``` **Test 4: Clarification Workflow (With Questions)** ```bash # MCP client call for generation task mcp-client call refine_prompt --prompt "Write a blog post" # Expected: Clarification needed response { "type": "clarification_needed", "task_type": "generation", "questions_for_ask_user_question": [ { "question": "Who is your target audience?", "header": "Q1", "multiSelect": false, "options": [ {"label": "Technical professionals", "description": "Technical professionals"}, {"label": "Business executives", "description": "Business executives"} ] } ], "answer_mapping": {"q0": "Who is your target audience?"} } ``` **Test 5: Complete Refinement with Answers** ```bash # MCP client call with answers from AskUserQuestion mcp-client call refine_prompt \ --prompt "Write a blog post" \ --answers '{"q0": "Technical professionals"}' \ --answer-mapping '{"q0": "Who is your target audience?"}' # Expected: Final refined prompt { "type": "refined", "prompt": "Write a comprehensive technical blog post about... targeted at software engineers...", "next_action": "This refined prompt is now ready to use..." } ``` ### MCP Tweak Tool Testing **Test 6: Prompt Modification** ```bash # MCP client call for prompt tweaking mcp-client call tweak_prompt \ --prompt "Write a technical blog post about Docker" \ --modification "make it more beginner-friendly" # Expected: Modified prompt { "type": "refined", "prompt": "Write an accessible beginner-friendly blog post about Docker..." } ``` ### MCP Error Handling Testing **Test 7: Missing Provider Configuration** ```bash # MCP client call without configured providers mcp-client call refine_prompt --prompt "Test prompt" # Expected: Configuration error { "type": "error", "error_type": "ConfigurationError", "message": "No provider configured. Please set API keys in environment." } ``` **Test 8: Invalid Provider** ```bash # MCP client call with invalid provider mcp-client call refine_prompt \ --prompt "Test prompt" \ --provider "invalid_provider" # Expected: Provider error { "type": "error", "error_type": "ConfigurationError", "message": "Provider 'invalid_provider' is not supported." } ``` ### MCP Integration Testing **Test 9: Model Discovery** ```bash # MCP client call to list models mcp-client call list_models --providers "google" --limit 5 # Expected: Model list with metadata { "type": "success", "providers": { "google": { "available": true, "models": [ {"id": "gemini-2.0-flash-exp", "name": "Gemini 2.0 Flash"}, {"id": "gemini-1.5-pro", "name": "Gemini 1.5 Pro"} ], "total_count": 15, "showing": 5 } } } ``` ### MCP Workflow Integration Examples **Example 1: AI Toolchain Integration** ```bash # Use MCP-refined prompt in larger AI workflow REFINED_PROMPT=$(mcp-client call refine_prompt --prompt "Create API documentation" | jq -r '.prompt') echo "$REFINED_PROMPT" | claude exec --generate-docs ``` **Example 2: Batch Processing with MCP** ```bash # Process multiple prompts through MCP for prompt in "Write a blog post" "Create a tutorial" "Draft an email"; do echo "Processing: $prompt" mcp-client call refine_prompt --prompt "$prompt" > "refined_${prompt// /_}.json" done ``` **Example 3: Interactive MCP Session** ```bash # Start MCP server for interactive use promptheus mcp # In another terminal, use MCP client for interactive refinement mcp-client interactive > refine_prompt "Write a technical article" # Handle clarification questions interactively # Use refined output with other tools ``` ### MCP Testing Checklist **Basic Functionality:** - [ ] MCP server starts without errors - [ ] All five tools are accessible - [ ] Provider configuration is detected - [ ] Error handling works correctly **Refinement Workflow:** - [ ] Direct refinement works for analysis tasks - [ ] Clarification workflow triggers for generation tasks - [ ] AskUserQuestion integration functions properly - [ ] Answer mapping works correctly **Integration Testing:** - [ ] MCP client can connect to server - [ ] JSON responses are properly formatted - [ ] Error responses include actionable information - [ ] Provider fallback behavior works **Performance Testing:** - [ ] Response times are reasonable (< 5 seconds) - [ ] No memory leaks during extended use - [ ] Concurrent requests are handled properly