Three Experts Prompt - README

Overview

The “Three Experts” prompt is a structured framework designed to help you get comprehensive, multi-perspective analysis on complex software engineering problems. By simulating three different experts, each with a distinct viewpoint, you’ll receive a more thorough analysis than you might get from a single perspective.

When to Use This Prompt

This prompt is particularly valuable when:

  • You’re facing architectural decisions with significant long-term implications
  • You’re unsure which of several approaches is optimal
  • You need to understand trade-offs between different solutions
  • You want to ensure you’ve considered edge cases and potential issues
  • You’re implementing a complex system and need structured thinking

How to Use the Prompt

  1. Copy the prompt: Copy the entire content from the PROMPT.md file.

  2. Fill in the tagged sections:
    • Replace the text in <task_description> with a detailed explanation of your problem
    • (Optional) Fill in <current_approach> with any existing implementation details
    • (Optional) Add any <specific_concerns> you want addressed

  3. Submit to an AI assistant: Paste the completed prompt into your preferred AI assistant (like Claude).

  4. Review the multi-perspective analysis: You’ll receive analysis from three different expert perspectives, followed by a collaborative recommendation.

Tips for Getting the Best Results

  • Be specific and detailed in your task description. Vague queries get vague responses.
  • Include relevant code snippets when applicable to give the experts concrete material to work with.
  • Mention your technology stack and any constraints you’re working under.
  • Specify performance requirements or other non-functional requirements that matter for your use case.
  • Ask follow-up questions if you need clarification on any part of the response.

Example Usage

Instead of asking:

“What’s the best way to implement caching in my app?”

Use the prompt with:

“ I need to implement a caching system for a Node.js e-commerce API that handles approximately 5,000 requests per minute. The data being cached includes product information, user preferences, and shopping cart contents. Currently, our database queries are creating a bottleneck, especially during peak hours. Cache invalidation is particularly challenging since product data is updated from multiple services.

We're currently using Redis for session management but not for data caching. We've experimented with in-memory caching using Node.js Map objects, but this doesn't work well in our clustered environment. - How should we handle cache invalidation across multiple services? - What's the appropriate TTL strategy for different types of data? - Should we use a write-through or write-behind caching pattern? - How do we monitor cache hit/miss rates?

”

Customization

Feel free to modify the experts’ roles based on your specific needs. For example, you might want:

  • A security expert
  • A DevOps specialist
  • A domain-specific expert (e.g., ML engineer, database specialist)
  • A UX-focused developer

You can adjust the prompt to specify exactly what type of expertise you need for your particular problem.